diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed0/default/resolved_config.yaml b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed0/default/resolved_config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9157fb8a7d06b8984a78717b67a8dd73003ca895 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed0/default/resolved_config.yaml @@ -0,0 +1,29 @@ +experiment_name: fewshot_39tasks_5shot_from_baseline26_processingline_seed0 +policy_type: baseline +base_model_path: ${HOME}/groot_robocasa/Atomic26_baseline/checkpoint-120000 +output_root: ${HOME}/groot_robocasa/Experiments_v2/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed0 +dataset: + dataset_soup: fewshot_heldout_atomic39_human_5shot +training: + num_gpus: 1 + batch_size: 64 + seed: 42 + filter_key_seed: 0 + manifest_path: ${HOME}/clvla/benchmarks/robocasa_v2/my_scripts/Fewshot/manifests/fewshot_heldout_atomic39/shot5_seed0.json + save_total_limit: 4 +phases: +- name: train_5shot_seed0_processingline_12k + max_steps: 12000 + save_steps: 0 + save_step_list: + - 2000 + - 5000 + - 8000 + - 12000 + trainable: + preset: processing_line_only + tune_llm: false + tune_visual: false + tune_projector: false + tune_diffusion_model: false +resolved_sweep: {} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed0/default/train_5shot_seed0_processingline_12k/checkpoint-2000/config.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed0/default/train_5shot_seed0_processingline_12k/checkpoint-2000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..47d7b77e4255263f3803a4b6f94ca6831e459b70 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed0/default/train_5shot_seed0_processingline_12k/checkpoint-2000/config.json @@ -0,0 +1,64 @@ +{ + "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" + ], + "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", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed0/default/train_5shot_seed0_processingline_12k/config.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed0/default/train_5shot_seed0_processingline_12k/config.json new file mode 100644 index 0000000000000000000000000000000000000000..47d7b77e4255263f3803a4b6f94ca6831e459b70 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed0/default/train_5shot_seed0_processingline_12k/config.json @@ -0,0 +1,64 @@ +{ + "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" + ], + "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", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/resolved_config.yaml b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/resolved_config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..2ab9aa7511119e421e5f4da7e35f5f9494451a49 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/resolved_config.yaml @@ -0,0 +1,29 @@ +experiment_name: fewshot_39tasks_5shot_from_baseline26_processingline_seed1 +policy_type: baseline +base_model_path: ${HOME}/groot_robocasa/Atomic26_baseline/checkpoint-120000 +output_root: ${HOME}/groot_robocasa/Experiments_v2/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1 +dataset: + dataset_soup: fewshot_heldout_atomic39_human_5shot +training: + num_gpus: 1 + batch_size: 64 + seed: 42 + filter_key_seed: 1 + manifest_path: ${HOME}/clvla/benchmarks/robocasa_v2/my_scripts/Fewshot/manifests/fewshot_heldout_atomic39/shot5_seed1.json + save_total_limit: 4 +phases: +- name: train_5shot_seed1_processingline_12k + max_steps: 12000 + save_steps: 0 + save_step_list: + - 2000 + - 5000 + - 8000 + - 12000 + trainable: + preset: processing_line_only + tune_llm: false + tune_visual: false + tune_projector: false + tune_diffusion_model: false +resolved_sweep: {} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/seed1.yaml b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/seed1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bbf6bafefd11295b1a23d52d4836e226ad513404 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/seed1.yaml @@ -0,0 +1,28 @@ +experiment_name: fewshot_39tasks_5shot_from_baseline26_processingline_seed1 +policy_type: baseline +base_model_path: ${HOME}/groot_robocasa/Atomic26_baseline/checkpoint-120000 +output_root: ${HOME}/groot_robocasa/Experiments_v2/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1 +dataset: + dataset_soup: fewshot_heldout_atomic39_human_5shot +training: + num_gpus: 1 + batch_size: 64 + seed: 42 + filter_key_seed: 1 + manifest_path: ${HOME}/clvla/benchmarks/robocasa_v2/my_scripts/Fewshot/manifests/fewshot_heldout_atomic39/shot5_seed1.json + save_total_limit: 4 +phases: +- name: train_5shot_seed1_processingline_12k + max_steps: 12000 + save_steps: 0 + save_step_list: + - 2000 + - 5000 + - 8000 + - 12000 + trainable: + preset: processing_line_only + tune_llm: false + tune_visual: false + tune_projector: false + tune_diffusion_model: false diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-12000/config.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-12000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..47d7b77e4255263f3803a4b6f94ca6831e459b70 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-12000/config.json @@ -0,0 +1,64 @@ +{ + "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" + ], + "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", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-12000/experiment_cfg/metadata.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-12000/experiment_cfg/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..27700cac5d292d4d42897b3c2e80bdc58e55c60e --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-12000/experiment_cfg/metadata.json @@ -0,0 +1,431 @@ +{ + "new_embodiment": { + "statistics": { + "state": { + "base_position": { + "max": [ + 7.778976917266846, + 0.33403685688972473, + 0.7223060727119446 + ], + "min": [ + -4.821934223175049, + -6.785110950469971, + 0.6994727253913879 + ], + "mean": [ + 2.5719731235192254, + -2.1489371002267235, + 0.7003612741268546 + ], + "std": [ + 1.669945884010784, + 1.3739411381361726, + 0.0009413282367754987 + ], + "q01": [ + -0.7944286339399775, + -5.486611518471242, + 0.699977555971739 + ], + "q99": [ + 6.019886905263648, + -0.5229184210026855, + 0.702930326282967 + ] + }, + "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, + 0.248543484335273, + 0.5959317405459531 + ], + "std": [ + 0.0, + 0.0, + 0.6655550240074092, + 0.374283995143559 + ], + "q01": [ + 0.0, + 0.0, + -1.0000000000000002, + 4.503254968910557e-06 + ], + "q99": [ + 0.0, + 0.0, + 1.0000000000000002, + 1.0000000000000002 + ] + }, + "end_effector_position_relative": { + "max": [ + 0.8848381638526917, + 0.8569459915161133, + 1.053884744644165 + ], + "min": [ + -0.5828549861907959, + -0.8486236333847046, + -0.3446122705936432 + ], + "mean": [ + 0.2465490874130862, + -0.004711658908481467, + 0.43463835464312073 + ], + "std": [ + 0.15798784670671595, + 0.25624102400497073, + 0.20836028727974915 + ], + "q01": [ + -0.020752051134640608, + -0.4837933887347875, + 0.1802942952289915 + ], + "q99": [ + 0.5386784714844007, + 0.4801424212024721, + 0.695095732975766 + ] + }, + "end_effector_rotation_relative": { + "max": [ + 0.9999949932098389, + 0.9997354745864868, + 0.9900747537612915, + 0.9735639095306396 + ], + "min": [ + -0.9999914169311523, + -0.9999884366989136, + -0.9975369572639465, + 6.754255821306288e-08 + ], + "mean": [ + -0.22531859714056926, + 0.046973702821026614, + -0.0568176967878771, + 0.12364307042955697 + ], + "std": [ + 0.8502144333612052, + 0.29591250318147405, + 0.30218053156837016, + 0.1636031353412931 + ], + "q01": [ + -0.9959057387172566, + -0.5665384745573048, + -0.5329334620834977, + 0.0014372279999397553 + ], + "q99": [ + 0.9904107951565839, + 0.6359623612021396, + 0.5026076022439377, + 0.42955559234879914 + ] + }, + "gripper_qpos": { + "max": [ + 0.05289580300450325, + 0.00937709677964449 + ], + "min": [ + -0.011669117026031017, + -0.06307756155729294 + ], + "mean": [ + 0.03188757060429838, + -0.0321793333086053 + ], + "std": [ + 0.01171052313737148, + 0.011427336801203189 + ], + "q01": [ + 0.006462901319459925, + -0.040557140599419166 + ], + "q99": [ + 0.04037619335388386, + -0.006523947835993347 + ] + } + }, + "action": { + "base_motion": { + "max": [ + 1.0, + 1.0, + 1.0, + 0.0 + ], + "min": [ + -1.0, + -1.0, + -1.0, + 0.0 + ], + "mean": [ + 0.006639007355857881, + 0.0027330421795561874, + -0.0017774441970214892, + 0.0 + ], + "std": [ + 0.11841528114040736, + 0.1041359083946544, + 0.08346265521208351, + 0.0 + ], + "q01": [ + -0.17562359675270237, + -0.21706402321529558, + -0.16273392585723104, + 0.0 + ], + "q99": [ + 0.2572918701644887, + 0.27022325575787115, + 0.1369626751049278, + 0.0 + ] + }, + "control_mode": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.9205539608927732 + ], + "std": [ + 0.39063657242534666 + ], + "q01": [ + -1.0 + ], + "q99": [ + -0.17711863866562935 + ] + }, + "end_effector_position": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + -0.00924803964621179, + 0.004053841761390575, + -0.08385115578460103 + ], + "std": [ + 0.45413072518223824, + 0.40158612714634856, + 0.41068634292849304 + ], + "q01": [ + -0.9669817145875343, + -0.9241307799783718, + -0.8264079503068825 + ], + "q99": [ + 0.8021511975722166, + 0.9241346582748478, + 0.8348116781502298 + ] + }, + "end_effector_rotation": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.007644157504353809, + -0.019962543093886335, + -0.002047110452602617 + ], + "std": [ + 0.11076012154791087, + 0.13176735140285795, + 0.12286528665895614 + ], + "q01": [ + -0.2659105439401192, + -0.3961694033813533, + -0.3434870832468936 + ], + "q99": [ + 0.3145786959266369, + 0.31309430361227547, + 0.34107754401942986 + ] + }, + "gripper_close": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.29870612267533575 + ], + "std": [ + 0.9543487555375691 + ], + "q01": [ + -1.0 + ], + "q99": [ + 0.7610674802484976 + ] + } + } + }, + "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/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-12000/model.safetensors.index.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-12000/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..5ef5f247b61eb93e89601c7da8334fc087e19750 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-12000/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/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-12000/trainer_state.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-12000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..746db732652f6d047274b69dc5ed43a32db9c4ee --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-12000/trainer_state.json @@ -0,0 +1,8434 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 20.100502512562816, + "eval_steps": 500, + "global_step": 12000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.01675041876046901, + "grad_norm": 1.1124801635742188, + "learning_rate": 4.5e-07, + "loss": 0.2576, + "step": 10 + }, + { + "epoch": 0.03350083752093802, + "grad_norm": 1.3764758110046387, + "learning_rate": 9.500000000000001e-07, + "loss": 0.2541, + "step": 20 + }, + { + "epoch": 0.05025125628140704, + "grad_norm": 1.0254155397415161, + "learning_rate": 1.45e-06, + "loss": 0.2369, + "step": 30 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 0.6008508801460266, + "learning_rate": 1.95e-06, + "loss": 0.2272, + "step": 40 + }, + { + "epoch": 0.08375209380234507, + "grad_norm": 1.085799217224121, + "learning_rate": 2.45e-06, + "loss": 0.2211, + "step": 50 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 0.9870124459266663, + "learning_rate": 2.9499999999999997e-06, + "loss": 0.2511, + "step": 60 + }, + { + "epoch": 0.11725293132328309, + "grad_norm": 0.9730237722396851, + "learning_rate": 3.4500000000000004e-06, + "loss": 0.2414, + "step": 70 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 0.792273998260498, + "learning_rate": 3.9499999999999995e-06, + "loss": 0.2311, + "step": 80 + }, + { + "epoch": 0.1507537688442211, + "grad_norm": 1.078325629234314, + "learning_rate": 4.450000000000001e-06, + "loss": 0.2185, + "step": 90 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 0.6237806081771851, + "learning_rate": 4.95e-06, + "loss": 0.1754, + "step": 100 + }, + { + "epoch": 0.18425460636515914, + "grad_norm": 0.7615986466407776, + "learning_rate": 5.45e-06, + "loss": 0.167, + "step": 110 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 0.6572969555854797, + "learning_rate": 5.95e-06, + "loss": 0.1568, + "step": 120 + }, + { + "epoch": 0.21775544388609716, + "grad_norm": 0.5472678542137146, + "learning_rate": 6.45e-06, + "loss": 0.1421, + "step": 130 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 0.4665081202983856, + "learning_rate": 6.95e-06, + "loss": 0.1526, + "step": 140 + }, + { + "epoch": 0.25125628140703515, + "grad_norm": 0.5054625272750854, + "learning_rate": 7.45e-06, + "loss": 0.1246, + "step": 150 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 0.5371837019920349, + "learning_rate": 7.95e-06, + "loss": 0.1179, + "step": 160 + }, + { + "epoch": 0.2847571189279732, + "grad_norm": 0.40661293268203735, + "learning_rate": 8.45e-06, + "loss": 0.1062, + "step": 170 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 0.2422139048576355, + "learning_rate": 8.95e-06, + "loss": 0.0963, + "step": 180 + }, + { + "epoch": 0.31825795644891125, + "grad_norm": 0.3034665584564209, + "learning_rate": 9.450000000000001e-06, + "loss": 0.0975, + "step": 190 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 0.2536943256855011, + "learning_rate": 9.95e-06, + "loss": 0.0934, + "step": 200 + }, + { + "epoch": 0.35175879396984927, + "grad_norm": 0.2668403685092926, + "learning_rate": 1.045e-05, + "loss": 0.0933, + "step": 210 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 0.22533497214317322, + "learning_rate": 1.095e-05, + "loss": 0.0792, + "step": 220 + }, + { + "epoch": 0.3852596314907873, + "grad_norm": 0.2040107697248459, + "learning_rate": 1.145e-05, + "loss": 0.0812, + "step": 230 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 0.19936227798461914, + "learning_rate": 1.1949999999999999e-05, + "loss": 0.0838, + "step": 240 + }, + { + "epoch": 0.4187604690117253, + "grad_norm": 0.2014681100845337, + "learning_rate": 1.245e-05, + "loss": 0.0804, + "step": 250 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 0.20712223649024963, + "learning_rate": 1.295e-05, + "loss": 0.0751, + "step": 260 + }, + { + "epoch": 0.45226130653266333, + "grad_norm": 0.23043803870677948, + "learning_rate": 1.345e-05, + "loss": 0.0747, + "step": 270 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 0.19819584488868713, + "learning_rate": 1.395e-05, + "loss": 0.0715, + "step": 280 + }, + { + "epoch": 0.48576214405360135, + "grad_norm": 0.2228541076183319, + "learning_rate": 1.445e-05, + "loss": 0.0708, + "step": 290 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 0.17452232539653778, + "learning_rate": 1.4950000000000001e-05, + "loss": 0.0685, + "step": 300 + }, + { + "epoch": 0.5192629815745393, + "grad_norm": 0.20126119256019592, + "learning_rate": 1.545e-05, + "loss": 0.0703, + "step": 310 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 0.2398366630077362, + "learning_rate": 1.5949999999999998e-05, + "loss": 0.0664, + "step": 320 + }, + { + "epoch": 0.5527638190954773, + "grad_norm": 0.21752075850963593, + "learning_rate": 1.645e-05, + "loss": 0.0679, + "step": 330 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 0.17013588547706604, + "learning_rate": 1.695e-05, + "loss": 0.0705, + "step": 340 + }, + { + "epoch": 0.5862646566164154, + "grad_norm": 0.2016950249671936, + "learning_rate": 1.745e-05, + "loss": 0.0687, + "step": 350 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 0.1707467883825302, + "learning_rate": 1.7950000000000003e-05, + "loss": 0.0675, + "step": 360 + }, + { + "epoch": 0.6197654941373534, + "grad_norm": 0.17180311679840088, + "learning_rate": 1.845e-05, + "loss": 0.0618, + "step": 370 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 0.21700109541416168, + "learning_rate": 1.8950000000000003e-05, + "loss": 0.0631, + "step": 380 + }, + { + "epoch": 0.6532663316582915, + "grad_norm": 0.21110156178474426, + "learning_rate": 1.945e-05, + "loss": 0.061, + "step": 390 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 0.17809593677520752, + "learning_rate": 1.995e-05, + "loss": 0.0585, + "step": 400 + }, + { + "epoch": 0.6867671691792295, + "grad_norm": 0.15697689354419708, + "learning_rate": 2.045e-05, + "loss": 0.0615, + "step": 410 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 0.17097905278205872, + "learning_rate": 2.095e-05, + "loss": 0.0568, + "step": 420 + }, + { + "epoch": 0.7202680067001676, + "grad_norm": 0.17003680765628815, + "learning_rate": 2.145e-05, + "loss": 0.0609, + "step": 430 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 0.20018595457077026, + "learning_rate": 2.195e-05, + "loss": 0.0538, + "step": 440 + }, + { + "epoch": 0.7537688442211056, + "grad_norm": 0.1872512400150299, + "learning_rate": 2.245e-05, + "loss": 0.0562, + "step": 450 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 0.210875004529953, + "learning_rate": 2.2950000000000002e-05, + "loss": 0.0538, + "step": 460 + }, + { + "epoch": 0.7872696817420436, + "grad_norm": 0.19456243515014648, + "learning_rate": 2.345e-05, + "loss": 0.0548, + "step": 470 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 0.19903069734573364, + "learning_rate": 2.395e-05, + "loss": 0.0536, + "step": 480 + }, + { + "epoch": 0.8207705192629816, + "grad_norm": 0.1998818963766098, + "learning_rate": 2.4449999999999998e-05, + "loss": 0.0533, + "step": 490 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 0.22372868657112122, + "learning_rate": 2.495e-05, + "loss": 0.0532, + "step": 500 + }, + { + "epoch": 0.8542713567839196, + "grad_norm": 0.2313748300075531, + "learning_rate": 2.5450000000000002e-05, + "loss": 0.049, + "step": 510 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 0.1595960259437561, + "learning_rate": 2.595e-05, + "loss": 0.0513, + "step": 520 + }, + { + "epoch": 0.8877721943048577, + "grad_norm": 0.3054298162460327, + "learning_rate": 2.6450000000000003e-05, + "loss": 0.0489, + "step": 530 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 0.15157485008239746, + "learning_rate": 2.695e-05, + "loss": 0.0501, + "step": 540 + }, + { + "epoch": 0.9212730318257957, + "grad_norm": 0.19291409850120544, + "learning_rate": 2.7450000000000003e-05, + "loss": 0.0428, + "step": 550 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 0.22616443037986755, + "learning_rate": 2.795e-05, + "loss": 0.048, + "step": 560 + }, + { + "epoch": 0.9547738693467337, + "grad_norm": 0.21530359983444214, + "learning_rate": 2.845e-05, + "loss": 0.051, + "step": 570 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 0.3027316629886627, + "learning_rate": 2.895e-05, + "loss": 0.0462, + "step": 580 + }, + { + "epoch": 0.9882747068676717, + "grad_norm": 0.16115690767765045, + "learning_rate": 2.945e-05, + "loss": 0.0438, + "step": 590 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 0.20157869160175323, + "learning_rate": 2.995e-05, + "loss": 0.0451, + "step": 600 + }, + { + "epoch": 1.0217755443886096, + "grad_norm": 0.19205620884895325, + "learning_rate": 2.9999953864407506e-05, + "loss": 0.0437, + "step": 610 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 0.1747637391090393, + "learning_rate": 2.99997943837114e-05, + "loss": 0.0454, + "step": 620 + }, + { + "epoch": 1.0552763819095476, + "grad_norm": 0.18132591247558594, + "learning_rate": 2.9999520989547326e-05, + "loss": 0.0488, + "step": 630 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 0.1378272920846939, + "learning_rate": 2.9999133683991525e-05, + "loss": 0.042, + "step": 640 + }, + { + "epoch": 1.0887772194304857, + "grad_norm": 0.17927886545658112, + "learning_rate": 2.9998632469985326e-05, + "loss": 0.0452, + "step": 650 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 0.1842069774866104, + "learning_rate": 2.999801735133512e-05, + "loss": 0.0418, + "step": 660 + }, + { + "epoch": 1.1222780569514237, + "grad_norm": 0.1894877851009369, + "learning_rate": 2.9997288332712325e-05, + "loss": 0.0421, + "step": 670 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 0.17502732574939728, + "learning_rate": 2.999644541965335e-05, + "loss": 0.0395, + "step": 680 + }, + { + "epoch": 1.1557788944723617, + "grad_norm": 0.19280853867530823, + "learning_rate": 2.9995488618559554e-05, + "loss": 0.0383, + "step": 690 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 0.21065080165863037, + "learning_rate": 2.9994417936697218e-05, + "loss": 0.0409, + "step": 700 + }, + { + "epoch": 1.1892797319932997, + "grad_norm": 0.15386411547660828, + "learning_rate": 2.9993233382197454e-05, + "loss": 0.0417, + "step": 710 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 0.18249723315238953, + "learning_rate": 2.999193496405616e-05, + "loss": 0.0386, + "step": 720 + }, + { + "epoch": 1.2227805695142377, + "grad_norm": 0.17197810113430023, + "learning_rate": 2.9990522692133962e-05, + "loss": 0.038, + "step": 730 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 0.15405157208442688, + "learning_rate": 2.9988996577156134e-05, + "loss": 0.0421, + "step": 740 + }, + { + "epoch": 1.2562814070351758, + "grad_norm": 0.16718165576457977, + "learning_rate": 2.9987356630712496e-05, + "loss": 0.04, + "step": 750 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 0.1574426144361496, + "learning_rate": 2.9985602865257354e-05, + "loss": 0.0403, + "step": 760 + }, + { + "epoch": 1.2897822445561138, + "grad_norm": 0.11435810476541519, + "learning_rate": 2.998373529410939e-05, + "loss": 0.0398, + "step": 770 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 0.16009581089019775, + "learning_rate": 2.9981753931451567e-05, + "loss": 0.037, + "step": 780 + }, + { + "epoch": 1.3232830820770518, + "grad_norm": 0.15292680263519287, + "learning_rate": 2.9979658792331016e-05, + "loss": 0.0375, + "step": 790 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 0.1592884510755539, + "learning_rate": 2.997744989265893e-05, + "loss": 0.037, + "step": 800 + }, + { + "epoch": 1.3567839195979898, + "grad_norm": 0.22414714097976685, + "learning_rate": 2.9975127249210426e-05, + "loss": 0.0344, + "step": 810 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 0.14030888676643372, + "learning_rate": 2.9972690879624445e-05, + "loss": 0.0364, + "step": 820 + }, + { + "epoch": 1.3902847571189278, + "grad_norm": 0.15426629781723022, + "learning_rate": 2.9970140802403587e-05, + "loss": 0.0368, + "step": 830 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 0.1377844214439392, + "learning_rate": 2.9967477036914002e-05, + "loss": 0.0365, + "step": 840 + }, + { + "epoch": 1.4237855946398659, + "grad_norm": 0.1314632147550583, + "learning_rate": 2.996469960338521e-05, + "loss": 0.0364, + "step": 850 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 0.1743793785572052, + "learning_rate": 2.9961808522909976e-05, + "loss": 0.0383, + "step": 860 + }, + { + "epoch": 1.457286432160804, + "grad_norm": 0.1816023737192154, + "learning_rate": 2.995880381744414e-05, + "loss": 0.0357, + "step": 870 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 0.19835101068019867, + "learning_rate": 2.995568550980643e-05, + "loss": 0.0383, + "step": 880 + }, + { + "epoch": 1.490787269681742, + "grad_norm": 0.13531947135925293, + "learning_rate": 2.9952453623678334e-05, + "loss": 0.0338, + "step": 890 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 0.13880357146263123, + "learning_rate": 2.9949108183603867e-05, + "loss": 0.036, + "step": 900 + }, + { + "epoch": 1.52428810720268, + "grad_norm": 0.18743862211704254, + "learning_rate": 2.994564921498943e-05, + "loss": 0.0358, + "step": 910 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 0.1355157345533371, + "learning_rate": 2.9942076744103594e-05, + "loss": 0.0342, + "step": 920 + }, + { + "epoch": 1.557788944723618, + "grad_norm": 0.16391365230083466, + "learning_rate": 2.9938390798076886e-05, + "loss": 0.0359, + "step": 930 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 0.1717372089624405, + "learning_rate": 2.993459140490163e-05, + "loss": 0.0344, + "step": 940 + }, + { + "epoch": 1.591289782244556, + "grad_norm": 0.19507038593292236, + "learning_rate": 2.9930678593431684e-05, + "loss": 0.0325, + "step": 950 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 0.14840291440486908, + "learning_rate": 2.9926652393382258e-05, + "loss": 0.0334, + "step": 960 + }, + { + "epoch": 1.624790619765494, + "grad_norm": 0.15482094883918762, + "learning_rate": 2.9922512835329655e-05, + "loss": 0.0338, + "step": 970 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 0.12408526986837387, + "learning_rate": 2.9918259950711073e-05, + "loss": 0.0324, + "step": 980 + }, + { + "epoch": 1.658291457286432, + "grad_norm": 0.1402246356010437, + "learning_rate": 2.991389377182435e-05, + "loss": 0.0363, + "step": 990 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 0.15740147233009338, + "learning_rate": 2.99094143318277e-05, + "loss": 0.0328, + "step": 1000 + }, + { + "epoch": 1.69179229480737, + "grad_norm": 0.14837925136089325, + "learning_rate": 2.9904821664739493e-05, + "loss": 0.0332, + "step": 1010 + }, + { + "epoch": 1.708542713567839, + "grad_norm": 0.1531188040971756, + "learning_rate": 2.990011580543799e-05, + "loss": 0.0326, + "step": 1020 + }, + { + "epoch": 1.725293132328308, + "grad_norm": 0.1466841697692871, + "learning_rate": 2.9895296789661065e-05, + "loss": 0.0326, + "step": 1030 + }, + { + "epoch": 1.742043551088777, + "grad_norm": 0.14380747079849243, + "learning_rate": 2.989036465400594e-05, + "loss": 0.0311, + "step": 1040 + }, + { + "epoch": 1.758793969849246, + "grad_norm": 0.15117715299129486, + "learning_rate": 2.9885319435928896e-05, + "loss": 0.0308, + "step": 1050 + }, + { + "epoch": 1.775544388609715, + "grad_norm": 0.1409190148115158, + "learning_rate": 2.9880161173745017e-05, + "loss": 0.0288, + "step": 1060 + }, + { + "epoch": 1.792294807370184, + "grad_norm": 0.11704489588737488, + "learning_rate": 2.9874889906627882e-05, + "loss": 0.0314, + "step": 1070 + }, + { + "epoch": 1.809045226130653, + "grad_norm": 0.1579095870256424, + "learning_rate": 2.9869505674609253e-05, + "loss": 0.0334, + "step": 1080 + }, + { + "epoch": 1.8257956448911221, + "grad_norm": 0.14541688561439514, + "learning_rate": 2.9864008518578794e-05, + "loss": 0.0301, + "step": 1090 + }, + { + "epoch": 1.8425460636515911, + "grad_norm": 0.11781132966279984, + "learning_rate": 2.9858398480283758e-05, + "loss": 0.031, + "step": 1100 + }, + { + "epoch": 1.8592964824120601, + "grad_norm": 0.1376485973596573, + "learning_rate": 2.9852675602328644e-05, + "loss": 0.0296, + "step": 1110 + }, + { + "epoch": 1.8760469011725294, + "grad_norm": 0.12675292789936066, + "learning_rate": 2.9846839928174923e-05, + "loss": 0.0289, + "step": 1120 + }, + { + "epoch": 1.8927973199329984, + "grad_norm": 0.14450091123580933, + "learning_rate": 2.984089150214065e-05, + "loss": 0.0297, + "step": 1130 + }, + { + "epoch": 1.9095477386934674, + "grad_norm": 0.15321670472621918, + "learning_rate": 2.9834830369400165e-05, + "loss": 0.0295, + "step": 1140 + }, + { + "epoch": 1.9262981574539364, + "grad_norm": 0.13381171226501465, + "learning_rate": 2.9828656575983747e-05, + "loss": 0.0279, + "step": 1150 + }, + { + "epoch": 1.9430485762144054, + "grad_norm": 0.20773489773273468, + "learning_rate": 2.9822370168777253e-05, + "loss": 0.0305, + "step": 1160 + }, + { + "epoch": 1.9597989949748744, + "grad_norm": 0.1649148315191269, + "learning_rate": 2.981597119552177e-05, + "loss": 0.0318, + "step": 1170 + }, + { + "epoch": 1.9765494137353434, + "grad_norm": 0.15318723022937775, + "learning_rate": 2.9809459704813236e-05, + "loss": 0.0283, + "step": 1180 + }, + { + "epoch": 1.9932998324958124, + "grad_norm": 0.11235921084880829, + "learning_rate": 2.980283574610211e-05, + "loss": 0.0281, + "step": 1190 + }, + { + "epoch": 2.0100502512562812, + "grad_norm": 0.14214928448200226, + "learning_rate": 2.9796099369692946e-05, + "loss": 0.0277, + "step": 1200 + }, + { + "epoch": 2.0268006700167502, + "grad_norm": 0.1358565390110016, + "learning_rate": 2.978925062674404e-05, + "loss": 0.0281, + "step": 1210 + }, + { + "epoch": 2.0435510887772192, + "grad_norm": 0.1451619565486908, + "learning_rate": 2.9782289569267057e-05, + "loss": 0.0279, + "step": 1220 + }, + { + "epoch": 2.0603015075376883, + "grad_norm": 0.1291060596704483, + "learning_rate": 2.977521625012659e-05, + "loss": 0.0307, + "step": 1230 + }, + { + "epoch": 2.0770519262981573, + "grad_norm": 0.13382935523986816, + "learning_rate": 2.9768030723039798e-05, + "loss": 0.0281, + "step": 1240 + }, + { + "epoch": 2.0938023450586263, + "grad_norm": 0.1389288753271103, + "learning_rate": 2.9760733042575984e-05, + "loss": 0.0286, + "step": 1250 + }, + { + "epoch": 2.1105527638190953, + "grad_norm": 0.13549239933490753, + "learning_rate": 2.9753323264156184e-05, + "loss": 0.0283, + "step": 1260 + }, + { + "epoch": 2.1273031825795643, + "grad_norm": 0.13530868291854858, + "learning_rate": 2.974580144405273e-05, + "loss": 0.0281, + "step": 1270 + }, + { + "epoch": 2.1440536013400333, + "grad_norm": 0.12069135159254074, + "learning_rate": 2.973816763938886e-05, + "loss": 0.0308, + "step": 1280 + }, + { + "epoch": 2.1608040201005023, + "grad_norm": 0.15732339024543762, + "learning_rate": 2.9730421908138236e-05, + "loss": 0.0274, + "step": 1290 + }, + { + "epoch": 2.1775544388609713, + "grad_norm": 0.14720749855041504, + "learning_rate": 2.9722564309124546e-05, + "loss": 0.0284, + "step": 1300 + }, + { + "epoch": 2.1943048576214403, + "grad_norm": 0.13191711902618408, + "learning_rate": 2.971459490202102e-05, + "loss": 0.0259, + "step": 1310 + }, + { + "epoch": 2.2110552763819094, + "grad_norm": 0.12703394889831543, + "learning_rate": 2.9706513747350017e-05, + "loss": 0.0245, + "step": 1320 + }, + { + "epoch": 2.2278056951423784, + "grad_norm": 0.13217298686504364, + "learning_rate": 2.9698320906482535e-05, + "loss": 0.0307, + "step": 1330 + }, + { + "epoch": 2.2445561139028474, + "grad_norm": 0.12877672910690308, + "learning_rate": 2.9690016441637747e-05, + "loss": 0.0276, + "step": 1340 + }, + { + "epoch": 2.2613065326633164, + "grad_norm": 0.13202929496765137, + "learning_rate": 2.9681600415882556e-05, + "loss": 0.0289, + "step": 1350 + }, + { + "epoch": 2.2780569514237854, + "grad_norm": 0.11203215271234512, + "learning_rate": 2.9673072893131068e-05, + "loss": 0.0274, + "step": 1360 + }, + { + "epoch": 2.2948073701842544, + "grad_norm": 0.09884139150381088, + "learning_rate": 2.9664433938144162e-05, + "loss": 0.0275, + "step": 1370 + }, + { + "epoch": 2.3115577889447234, + "grad_norm": 0.12471596896648407, + "learning_rate": 2.9655683616528958e-05, + "loss": 0.0252, + "step": 1380 + }, + { + "epoch": 2.3283082077051924, + "grad_norm": 0.15619947016239166, + "learning_rate": 2.9646821994738322e-05, + "loss": 0.0267, + "step": 1390 + }, + { + "epoch": 2.3450586264656614, + "grad_norm": 0.11800948530435562, + "learning_rate": 2.9637849140070393e-05, + "loss": 0.025, + "step": 1400 + }, + { + "epoch": 2.3618090452261304, + "grad_norm": 0.12330026179552078, + "learning_rate": 2.9628765120668033e-05, + "loss": 0.0242, + "step": 1410 + }, + { + "epoch": 2.3785594639865995, + "grad_norm": 0.14527784287929535, + "learning_rate": 2.9619570005518336e-05, + "loss": 0.0258, + "step": 1420 + }, + { + "epoch": 2.3953098827470685, + "grad_norm": 0.1384161412715912, + "learning_rate": 2.9610263864452077e-05, + "loss": 0.0245, + "step": 1430 + }, + { + "epoch": 2.4120603015075375, + "grad_norm": 0.14128261804580688, + "learning_rate": 2.9600846768143225e-05, + "loss": 0.0243, + "step": 1440 + }, + { + "epoch": 2.4288107202680065, + "grad_norm": 0.12654323875904083, + "learning_rate": 2.959131878810836e-05, + "loss": 0.0257, + "step": 1450 + }, + { + "epoch": 2.4455611390284755, + "grad_norm": 0.19983522593975067, + "learning_rate": 2.9581679996706157e-05, + "loss": 0.0278, + "step": 1460 + }, + { + "epoch": 2.4623115577889445, + "grad_norm": 0.14568129181861877, + "learning_rate": 2.9571930467136833e-05, + "loss": 0.0256, + "step": 1470 + }, + { + "epoch": 2.4790619765494135, + "grad_norm": 0.12198968231678009, + "learning_rate": 2.956207027344159e-05, + "loss": 0.0257, + "step": 1480 + }, + { + "epoch": 2.4958123953098825, + "grad_norm": 0.11267124861478806, + "learning_rate": 2.9552099490502026e-05, + "loss": 0.0246, + "step": 1490 + }, + { + "epoch": 2.5125628140703515, + "grad_norm": 0.11769300699234009, + "learning_rate": 2.9542018194039634e-05, + "loss": 0.0249, + "step": 1500 + }, + { + "epoch": 2.5293132328308205, + "grad_norm": 0.11569931358098984, + "learning_rate": 2.953182646061514e-05, + "loss": 0.0278, + "step": 1510 + }, + { + "epoch": 2.5460636515912896, + "grad_norm": 0.12251350283622742, + "learning_rate": 2.9521524367627994e-05, + "loss": 0.0233, + "step": 1520 + }, + { + "epoch": 2.5628140703517586, + "grad_norm": 0.11296738684177399, + "learning_rate": 2.9511111993315743e-05, + "loss": 0.0251, + "step": 1530 + }, + { + "epoch": 2.5795644891122276, + "grad_norm": 0.11284096539020538, + "learning_rate": 2.9500589416753447e-05, + "loss": 0.0261, + "step": 1540 + }, + { + "epoch": 2.5963149078726966, + "grad_norm": 0.14394700527191162, + "learning_rate": 2.9489956717853088e-05, + "loss": 0.0272, + "step": 1550 + }, + { + "epoch": 2.6130653266331656, + "grad_norm": 0.12496906518936157, + "learning_rate": 2.9479213977362935e-05, + "loss": 0.0255, + "step": 1560 + }, + { + "epoch": 2.6298157453936346, + "grad_norm": 0.1484213024377823, + "learning_rate": 2.9468361276866973e-05, + "loss": 0.0257, + "step": 1570 + }, + { + "epoch": 2.6465661641541036, + "grad_norm": 0.13880600035190582, + "learning_rate": 2.9457398698784243e-05, + "loss": 0.0226, + "step": 1580 + }, + { + "epoch": 2.6633165829145726, + "grad_norm": 0.1260875016450882, + "learning_rate": 2.944632632636823e-05, + "loss": 0.026, + "step": 1590 + }, + { + "epoch": 2.6800670016750416, + "grad_norm": 0.14514032006263733, + "learning_rate": 2.9435144243706255e-05, + "loss": 0.0251, + "step": 1600 + }, + { + "epoch": 2.6968174204355106, + "grad_norm": 0.10520205646753311, + "learning_rate": 2.9423852535718805e-05, + "loss": 0.023, + "step": 1610 + }, + { + "epoch": 2.7135678391959797, + "grad_norm": 0.09249555319547653, + "learning_rate": 2.9412451288158883e-05, + "loss": 0.0238, + "step": 1620 + }, + { + "epoch": 2.7303182579564487, + "grad_norm": 0.08644948154687881, + "learning_rate": 2.94009405876114e-05, + "loss": 0.0242, + "step": 1630 + }, + { + "epoch": 2.7470686767169177, + "grad_norm": 0.12189530581235886, + "learning_rate": 2.9389320521492464e-05, + "loss": 0.0216, + "step": 1640 + }, + { + "epoch": 2.7638190954773867, + "grad_norm": 0.10813789814710617, + "learning_rate": 2.9377591178048765e-05, + "loss": 0.0249, + "step": 1650 + }, + { + "epoch": 2.7805695142378557, + "grad_norm": 0.10432349145412445, + "learning_rate": 2.9365752646356857e-05, + "loss": 0.0216, + "step": 1660 + }, + { + "epoch": 2.7973199329983247, + "grad_norm": 0.08932849764823914, + "learning_rate": 2.935380501632253e-05, + "loss": 0.0245, + "step": 1670 + }, + { + "epoch": 2.8140703517587937, + "grad_norm": 0.12938445806503296, + "learning_rate": 2.934174837868008e-05, + "loss": 0.0252, + "step": 1680 + }, + { + "epoch": 2.8308207705192627, + "grad_norm": 0.20361070334911346, + "learning_rate": 2.932958282499167e-05, + "loss": 0.0258, + "step": 1690 + }, + { + "epoch": 2.8475711892797317, + "grad_norm": 0.12765756249427795, + "learning_rate": 2.9317308447646585e-05, + "loss": 0.0256, + "step": 1700 + }, + { + "epoch": 2.8643216080402008, + "grad_norm": 0.13113832473754883, + "learning_rate": 2.9304925339860565e-05, + "loss": 0.0258, + "step": 1710 + }, + { + "epoch": 2.8810720268006698, + "grad_norm": 0.13574381172657013, + "learning_rate": 2.929243359567509e-05, + "loss": 0.0231, + "step": 1720 + }, + { + "epoch": 2.8978224455611388, + "grad_norm": 0.15551957488059998, + "learning_rate": 2.9279833309956637e-05, + "loss": 0.0233, + "step": 1730 + }, + { + "epoch": 2.914572864321608, + "grad_norm": 0.1148996576666832, + "learning_rate": 2.9267124578396013e-05, + "loss": 0.0215, + "step": 1740 + }, + { + "epoch": 2.931323283082077, + "grad_norm": 0.1132749691605568, + "learning_rate": 2.9254307497507584e-05, + "loss": 0.0255, + "step": 1750 + }, + { + "epoch": 2.948073701842546, + "grad_norm": 0.12405268102884293, + "learning_rate": 2.924138216462856e-05, + "loss": 0.0232, + "step": 1760 + }, + { + "epoch": 2.964824120603015, + "grad_norm": 0.13782398402690887, + "learning_rate": 2.922834867791825e-05, + "loss": 0.0257, + "step": 1770 + }, + { + "epoch": 2.981574539363484, + "grad_norm": 0.13731901347637177, + "learning_rate": 2.9215207136357322e-05, + "loss": 0.0235, + "step": 1780 + }, + { + "epoch": 2.998324958123953, + "grad_norm": 0.11207369714975357, + "learning_rate": 2.9201957639747056e-05, + "loss": 0.0254, + "step": 1790 + }, + { + "epoch": 3.0150753768844223, + "grad_norm": 0.12427914887666702, + "learning_rate": 2.9188600288708565e-05, + "loss": 0.0231, + "step": 1800 + }, + { + "epoch": 3.0318257956448913, + "grad_norm": 0.12067142128944397, + "learning_rate": 2.9175135184682053e-05, + "loss": 0.0211, + "step": 1810 + }, + { + "epoch": 3.0485762144053603, + "grad_norm": 0.1151166558265686, + "learning_rate": 2.9161562429926038e-05, + "loss": 0.0228, + "step": 1820 + }, + { + "epoch": 3.0653266331658293, + "grad_norm": 0.12743020057678223, + "learning_rate": 2.9147882127516568e-05, + "loss": 0.0227, + "step": 1830 + }, + { + "epoch": 3.0820770519262983, + "grad_norm": 0.12181981652975082, + "learning_rate": 2.913409438134644e-05, + "loss": 0.024, + "step": 1840 + }, + { + "epoch": 3.0988274706867673, + "grad_norm": 0.1195250153541565, + "learning_rate": 2.9120199296124436e-05, + "loss": 0.0251, + "step": 1850 + }, + { + "epoch": 3.1155778894472363, + "grad_norm": 0.09719762206077576, + "learning_rate": 2.9106196977374477e-05, + "loss": 0.0232, + "step": 1860 + }, + { + "epoch": 3.1323283082077054, + "grad_norm": 0.11931390315294266, + "learning_rate": 2.9092087531434878e-05, + "loss": 0.0236, + "step": 1870 + }, + { + "epoch": 3.1490787269681744, + "grad_norm": 0.10895305871963501, + "learning_rate": 2.9077871065457495e-05, + "loss": 0.023, + "step": 1880 + }, + { + "epoch": 3.1658291457286434, + "grad_norm": 0.1338292807340622, + "learning_rate": 2.9063547687406945e-05, + "loss": 0.024, + "step": 1890 + }, + { + "epoch": 3.1825795644891124, + "grad_norm": 0.11511155962944031, + "learning_rate": 2.9049117506059756e-05, + "loss": 0.0208, + "step": 1900 + }, + { + "epoch": 3.1993299832495814, + "grad_norm": 0.11553452163934708, + "learning_rate": 2.9034580631003564e-05, + "loss": 0.0232, + "step": 1910 + }, + { + "epoch": 3.2160804020100504, + "grad_norm": 0.1008048877120018, + "learning_rate": 2.9019937172636276e-05, + "loss": 0.0222, + "step": 1920 + }, + { + "epoch": 3.2328308207705194, + "grad_norm": 0.10765612125396729, + "learning_rate": 2.9005187242165228e-05, + "loss": 0.0226, + "step": 1930 + }, + { + "epoch": 3.2495812395309884, + "grad_norm": 0.11185599863529205, + "learning_rate": 2.899033095160633e-05, + "loss": 0.0213, + "step": 1940 + }, + { + "epoch": 3.2663316582914574, + "grad_norm": 0.09481639415025711, + "learning_rate": 2.8975368413783235e-05, + "loss": 0.0217, + "step": 1950 + }, + { + "epoch": 3.2830820770519265, + "grad_norm": 0.11557790637016296, + "learning_rate": 2.8960299742326473e-05, + "loss": 0.0231, + "step": 1960 + }, + { + "epoch": 3.2998324958123955, + "grad_norm": 0.13437284529209137, + "learning_rate": 2.8945125051672595e-05, + "loss": 0.0234, + "step": 1970 + }, + { + "epoch": 3.3165829145728645, + "grad_norm": 0.11012782156467438, + "learning_rate": 2.892984445706328e-05, + "loss": 0.0225, + "step": 1980 + }, + { + "epoch": 3.3333333333333335, + "grad_norm": 0.11317654699087143, + "learning_rate": 2.891445807454449e-05, + "loss": 0.024, + "step": 1990 + }, + { + "epoch": 3.3500837520938025, + "grad_norm": 0.19488593935966492, + "learning_rate": 2.8898966020965576e-05, + "loss": 0.024, + "step": 2000 + }, + { + "epoch": 3.3668341708542715, + "grad_norm": 0.10659220814704895, + "learning_rate": 2.8883368413978386e-05, + "loss": 0.0225, + "step": 2010 + }, + { + "epoch": 3.3835845896147405, + "grad_norm": 0.09751500934362411, + "learning_rate": 2.8867665372036375e-05, + "loss": 0.0258, + "step": 2020 + }, + { + "epoch": 3.4003350083752095, + "grad_norm": 0.09632577747106552, + "learning_rate": 2.8851857014393705e-05, + "loss": 0.0222, + "step": 2030 + }, + { + "epoch": 3.4170854271356785, + "grad_norm": 0.12793496251106262, + "learning_rate": 2.8835943461104354e-05, + "loss": 0.0214, + "step": 2040 + }, + { + "epoch": 3.4338358458961475, + "grad_norm": 0.11942226439714432, + "learning_rate": 2.8819924833021176e-05, + "loss": 0.023, + "step": 2050 + }, + { + "epoch": 3.4505862646566166, + "grad_norm": 0.10179294645786285, + "learning_rate": 2.8803801251795007e-05, + "loss": 0.0212, + "step": 2060 + }, + { + "epoch": 3.4673366834170856, + "grad_norm": 0.1228783056139946, + "learning_rate": 2.8787572839873725e-05, + "loss": 0.0225, + "step": 2070 + }, + { + "epoch": 3.4840871021775546, + "grad_norm": 0.13721901178359985, + "learning_rate": 2.8771239720501333e-05, + "loss": 0.0199, + "step": 2080 + }, + { + "epoch": 3.5008375209380236, + "grad_norm": 0.1087748259305954, + "learning_rate": 2.8754802017717015e-05, + "loss": 0.0228, + "step": 2090 + }, + { + "epoch": 3.5175879396984926, + "grad_norm": 0.14415811002254486, + "learning_rate": 2.87382598563542e-05, + "loss": 0.0241, + "step": 2100 + }, + { + "epoch": 3.5343383584589616, + "grad_norm": 0.10477285832166672, + "learning_rate": 2.8721613362039607e-05, + "loss": 0.02, + "step": 2110 + }, + { + "epoch": 3.5510887772194306, + "grad_norm": 0.10786721855401993, + "learning_rate": 2.8704862661192295e-05, + "loss": 0.0212, + "step": 2120 + }, + { + "epoch": 3.5678391959798996, + "grad_norm": 0.10725842416286469, + "learning_rate": 2.86880078810227e-05, + "loss": 0.022, + "step": 2130 + }, + { + "epoch": 3.5845896147403686, + "grad_norm": 0.11172711849212646, + "learning_rate": 2.8671049149531677e-05, + "loss": 0.021, + "step": 2140 + }, + { + "epoch": 3.6013400335008376, + "grad_norm": 0.10756686329841614, + "learning_rate": 2.8653986595509524e-05, + "loss": 0.0208, + "step": 2150 + }, + { + "epoch": 3.6180904522613067, + "grad_norm": 0.08842871338129044, + "learning_rate": 2.8636820348534987e-05, + "loss": 0.0208, + "step": 2160 + }, + { + "epoch": 3.6348408710217757, + "grad_norm": 0.10979699343442917, + "learning_rate": 2.861955053897431e-05, + "loss": 0.0204, + "step": 2170 + }, + { + "epoch": 3.6515912897822447, + "grad_norm": 0.0892506092786789, + "learning_rate": 2.8602177297980213e-05, + "loss": 0.0209, + "step": 2180 + }, + { + "epoch": 3.6683417085427137, + "grad_norm": 0.10864245146512985, + "learning_rate": 2.8584700757490913e-05, + "loss": 0.0214, + "step": 2190 + }, + { + "epoch": 3.6850921273031827, + "grad_norm": 0.12509699165821075, + "learning_rate": 2.8567121050229122e-05, + "loss": 0.0236, + "step": 2200 + }, + { + "epoch": 3.7018425460636517, + "grad_norm": 0.11275574564933777, + "learning_rate": 2.8549438309701028e-05, + "loss": 0.0237, + "step": 2210 + }, + { + "epoch": 3.7185929648241207, + "grad_norm": 0.12549647688865662, + "learning_rate": 2.8531652670195292e-05, + "loss": 0.021, + "step": 2220 + }, + { + "epoch": 3.7353433835845897, + "grad_norm": 0.11705254763364792, + "learning_rate": 2.8513764266782025e-05, + "loss": 0.0224, + "step": 2230 + }, + { + "epoch": 3.7520938023450587, + "grad_norm": 0.16595874726772308, + "learning_rate": 2.849577323531176e-05, + "loss": 0.0224, + "step": 2240 + }, + { + "epoch": 3.7688442211055277, + "grad_norm": 0.10253427922725677, + "learning_rate": 2.8477679712414424e-05, + "loss": 0.0215, + "step": 2250 + }, + { + "epoch": 3.7855946398659968, + "grad_norm": 0.10289912670850754, + "learning_rate": 2.8459483835498298e-05, + "loss": 0.0237, + "step": 2260 + }, + { + "epoch": 3.8023450586264658, + "grad_norm": 0.11400792002677917, + "learning_rate": 2.844118574274896e-05, + "loss": 0.0205, + "step": 2270 + }, + { + "epoch": 3.819095477386935, + "grad_norm": 0.11081373691558838, + "learning_rate": 2.842278557312828e-05, + "loss": 0.0184, + "step": 2280 + }, + { + "epoch": 3.835845896147404, + "grad_norm": 0.13571597635746002, + "learning_rate": 2.84042834663733e-05, + "loss": 0.0207, + "step": 2290 + }, + { + "epoch": 3.852596314907873, + "grad_norm": 0.10395380854606628, + "learning_rate": 2.8385679562995235e-05, + "loss": 0.0209, + "step": 2300 + }, + { + "epoch": 3.869346733668342, + "grad_norm": 0.10459823906421661, + "learning_rate": 2.8366974004278357e-05, + "loss": 0.0242, + "step": 2310 + }, + { + "epoch": 3.886097152428811, + "grad_norm": 0.10516241937875748, + "learning_rate": 2.834816693227895e-05, + "loss": 0.0205, + "step": 2320 + }, + { + "epoch": 3.90284757118928, + "grad_norm": 0.1241426020860672, + "learning_rate": 2.8329258489824234e-05, + "loss": 0.0227, + "step": 2330 + }, + { + "epoch": 3.919597989949749, + "grad_norm": 0.10768191516399384, + "learning_rate": 2.8310248820511248e-05, + "loss": 0.0215, + "step": 2340 + }, + { + "epoch": 3.936348408710218, + "grad_norm": 0.10944578796625137, + "learning_rate": 2.829113806870581e-05, + "loss": 0.0214, + "step": 2350 + }, + { + "epoch": 3.953098827470687, + "grad_norm": 0.11351463198661804, + "learning_rate": 2.8271926379541363e-05, + "loss": 0.022, + "step": 2360 + }, + { + "epoch": 3.969849246231156, + "grad_norm": 0.12451263517141342, + "learning_rate": 2.8252613898917923e-05, + "loss": 0.0216, + "step": 2370 + }, + { + "epoch": 3.986599664991625, + "grad_norm": 0.11138986051082611, + "learning_rate": 2.8233200773500942e-05, + "loss": 0.0215, + "step": 2380 + }, + { + "epoch": 4.0033500837520934, + "grad_norm": 0.14621508121490479, + "learning_rate": 2.8213687150720204e-05, + "loss": 0.0227, + "step": 2390 + }, + { + "epoch": 4.0201005025125625, + "grad_norm": 0.12678438425064087, + "learning_rate": 2.8194073178768703e-05, + "loss": 0.0211, + "step": 2400 + }, + { + "epoch": 4.0368509212730315, + "grad_norm": 0.10121660679578781, + "learning_rate": 2.817435900660152e-05, + "loss": 0.0191, + "step": 2410 + }, + { + "epoch": 4.0536013400335005, + "grad_norm": 0.09031516313552856, + "learning_rate": 2.8154544783934688e-05, + "loss": 0.0241, + "step": 2420 + }, + { + "epoch": 4.0703517587939695, + "grad_norm": 0.12526579201221466, + "learning_rate": 2.8134630661244055e-05, + "loss": 0.0212, + "step": 2430 + }, + { + "epoch": 4.0871021775544385, + "grad_norm": 0.09737418591976166, + "learning_rate": 2.8114616789764146e-05, + "loss": 0.0196, + "step": 2440 + }, + { + "epoch": 4.1038525963149075, + "grad_norm": 0.12097261101007462, + "learning_rate": 2.8094503321487017e-05, + "loss": 0.0208, + "step": 2450 + }, + { + "epoch": 4.1206030150753765, + "grad_norm": 0.10906875133514404, + "learning_rate": 2.8074290409161084e-05, + "loss": 0.0183, + "step": 2460 + }, + { + "epoch": 4.1373534338358455, + "grad_norm": 0.12879160046577454, + "learning_rate": 2.8053978206289984e-05, + "loss": 0.0194, + "step": 2470 + }, + { + "epoch": 4.1541038525963145, + "grad_norm": 0.0820513367652893, + "learning_rate": 2.803356686713139e-05, + "loss": 0.0214, + "step": 2480 + }, + { + "epoch": 4.1708542713567835, + "grad_norm": 0.07784825563430786, + "learning_rate": 2.801305654669587e-05, + "loss": 0.0217, + "step": 2490 + }, + { + "epoch": 4.187604690117253, + "grad_norm": 0.0847000777721405, + "learning_rate": 2.799244740074566e-05, + "loss": 0.0195, + "step": 2500 + }, + { + "epoch": 4.204355108877722, + "grad_norm": 0.10731722414493561, + "learning_rate": 2.797173958579353e-05, + "loss": 0.0206, + "step": 2510 + }, + { + "epoch": 4.221105527638191, + "grad_norm": 0.1059015691280365, + "learning_rate": 2.7950933259101575e-05, + "loss": 0.0189, + "step": 2520 + }, + { + "epoch": 4.23785594639866, + "grad_norm": 0.10778709501028061, + "learning_rate": 2.793002857868001e-05, + "loss": 0.0208, + "step": 2530 + }, + { + "epoch": 4.254606365159129, + "grad_norm": 0.08753228187561035, + "learning_rate": 2.7909025703285983e-05, + "loss": 0.0195, + "step": 2540 + }, + { + "epoch": 4.271356783919598, + "grad_norm": 0.1175403892993927, + "learning_rate": 2.7887924792422385e-05, + "loss": 0.023, + "step": 2550 + }, + { + "epoch": 4.288107202680067, + "grad_norm": 0.12728635966777802, + "learning_rate": 2.78667260063366e-05, + "loss": 0.0215, + "step": 2560 + }, + { + "epoch": 4.304857621440536, + "grad_norm": 0.10965393483638763, + "learning_rate": 2.784542950601931e-05, + "loss": 0.019, + "step": 2570 + }, + { + "epoch": 4.321608040201005, + "grad_norm": 0.10030016303062439, + "learning_rate": 2.7824035453203293e-05, + "loss": 0.0214, + "step": 2580 + }, + { + "epoch": 4.338358458961474, + "grad_norm": 0.10159602761268616, + "learning_rate": 2.7802544010362154e-05, + "loss": 0.018, + "step": 2590 + }, + { + "epoch": 4.355108877721943, + "grad_norm": 0.08524280786514282, + "learning_rate": 2.778095534070912e-05, + "loss": 0.0225, + "step": 2600 + }, + { + "epoch": 4.371859296482412, + "grad_norm": 0.08864463865756989, + "learning_rate": 2.7759269608195777e-05, + "loss": 0.0214, + "step": 2610 + }, + { + "epoch": 4.388609715242881, + "grad_norm": 0.0836586281657219, + "learning_rate": 2.773748697751086e-05, + "loss": 0.019, + "step": 2620 + }, + { + "epoch": 4.40536013400335, + "grad_norm": 0.10019984096288681, + "learning_rate": 2.7715607614078956e-05, + "loss": 0.0204, + "step": 2630 + }, + { + "epoch": 4.422110552763819, + "grad_norm": 0.097859688103199, + "learning_rate": 2.7693631684059305e-05, + "loss": 0.0217, + "step": 2640 + }, + { + "epoch": 4.438860971524288, + "grad_norm": 0.09158801287412643, + "learning_rate": 2.767155935434448e-05, + "loss": 0.0197, + "step": 2650 + }, + { + "epoch": 4.455611390284757, + "grad_norm": 0.09695200622081757, + "learning_rate": 2.7649390792559146e-05, + "loss": 0.0206, + "step": 2660 + }, + { + "epoch": 4.472361809045226, + "grad_norm": 0.09822610765695572, + "learning_rate": 2.7627126167058807e-05, + "loss": 0.0203, + "step": 2670 + }, + { + "epoch": 4.489112227805695, + "grad_norm": 0.09746448695659637, + "learning_rate": 2.7604765646928482e-05, + "loss": 0.0204, + "step": 2680 + }, + { + "epoch": 4.505862646566165, + "grad_norm": 0.10244615375995636, + "learning_rate": 2.758230940198147e-05, + "loss": 0.0215, + "step": 2690 + }, + { + "epoch": 4.522613065326633, + "grad_norm": 0.10013485699892044, + "learning_rate": 2.755975760275802e-05, + "loss": 0.0198, + "step": 2700 + }, + { + "epoch": 4.539363484087103, + "grad_norm": 0.09892448037862778, + "learning_rate": 2.7537110420524057e-05, + "loss": 0.0199, + "step": 2710 + }, + { + "epoch": 4.556113902847571, + "grad_norm": 0.09825485199689865, + "learning_rate": 2.7514368027269875e-05, + "loss": 0.0186, + "step": 2720 + }, + { + "epoch": 4.572864321608041, + "grad_norm": 0.11656782776117325, + "learning_rate": 2.749153059570883e-05, + "loss": 0.0217, + "step": 2730 + }, + { + "epoch": 4.589614740368509, + "grad_norm": 0.08932335674762726, + "learning_rate": 2.7468598299276044e-05, + "loss": 0.0197, + "step": 2740 + }, + { + "epoch": 4.606365159128979, + "grad_norm": 0.09198352694511414, + "learning_rate": 2.7445571312127058e-05, + "loss": 0.0168, + "step": 2750 + }, + { + "epoch": 4.623115577889447, + "grad_norm": 0.09846565872430801, + "learning_rate": 2.7422449809136538e-05, + "loss": 0.0183, + "step": 2760 + }, + { + "epoch": 4.639865996649917, + "grad_norm": 0.07488419115543365, + "learning_rate": 2.739923396589693e-05, + "loss": 0.0209, + "step": 2770 + }, + { + "epoch": 4.656616415410385, + "grad_norm": 0.11510682106018066, + "learning_rate": 2.737592395871713e-05, + "loss": 0.0184, + "step": 2780 + }, + { + "epoch": 4.673366834170855, + "grad_norm": 0.08309126645326614, + "learning_rate": 2.7352519964621146e-05, + "loss": 0.0175, + "step": 2790 + }, + { + "epoch": 4.690117252931323, + "grad_norm": 0.0989244133234024, + "learning_rate": 2.7329022161346758e-05, + "loss": 0.02, + "step": 2800 + }, + { + "epoch": 4.706867671691793, + "grad_norm": 0.09160429984331131, + "learning_rate": 2.730543072734417e-05, + "loss": 0.0211, + "step": 2810 + }, + { + "epoch": 4.723618090452261, + "grad_norm": 0.13737095892429352, + "learning_rate": 2.728174584177463e-05, + "loss": 0.02, + "step": 2820 + }, + { + "epoch": 4.740368509212731, + "grad_norm": 0.09824742376804352, + "learning_rate": 2.725796768450911e-05, + "loss": 0.0187, + "step": 2830 + }, + { + "epoch": 4.757118927973199, + "grad_norm": 0.07674989849328995, + "learning_rate": 2.7234096436126903e-05, + "loss": 0.0201, + "step": 2840 + }, + { + "epoch": 4.773869346733669, + "grad_norm": 0.09987271577119827, + "learning_rate": 2.721013227791428e-05, + "loss": 0.0201, + "step": 2850 + }, + { + "epoch": 4.790619765494137, + "grad_norm": 0.10610336065292358, + "learning_rate": 2.7186075391863088e-05, + "loss": 0.0216, + "step": 2860 + }, + { + "epoch": 4.807370184254607, + "grad_norm": 0.09385973960161209, + "learning_rate": 2.7161925960669383e-05, + "loss": 0.0218, + "step": 2870 + }, + { + "epoch": 4.824120603015075, + "grad_norm": 0.1261572688817978, + "learning_rate": 2.713768416773205e-05, + "loss": 0.0201, + "step": 2880 + }, + { + "epoch": 4.840871021775545, + "grad_norm": 0.06238450109958649, + "learning_rate": 2.7113350197151394e-05, + "loss": 0.0192, + "step": 2890 + }, + { + "epoch": 4.857621440536013, + "grad_norm": 0.08462442457675934, + "learning_rate": 2.708892423372774e-05, + "loss": 0.0189, + "step": 2900 + }, + { + "epoch": 4.874371859296483, + "grad_norm": 0.11602508276700974, + "learning_rate": 2.7064406462960044e-05, + "loss": 0.0175, + "step": 2910 + }, + { + "epoch": 4.891122278056951, + "grad_norm": 0.08591975271701813, + "learning_rate": 2.7039797071044482e-05, + "loss": 0.023, + "step": 2920 + }, + { + "epoch": 4.907872696817421, + "grad_norm": 0.11215902119874954, + "learning_rate": 2.7015096244873028e-05, + "loss": 0.0204, + "step": 2930 + }, + { + "epoch": 4.924623115577889, + "grad_norm": 0.11343713849782944, + "learning_rate": 2.6990304172032038e-05, + "loss": 0.0199, + "step": 2940 + }, + { + "epoch": 4.941373534338359, + "grad_norm": 0.08644749969244003, + "learning_rate": 2.6965421040800823e-05, + "loss": 0.0219, + "step": 2950 + }, + { + "epoch": 4.958123953098827, + "grad_norm": 0.0927010178565979, + "learning_rate": 2.6940447040150228e-05, + "loss": 0.0194, + "step": 2960 + }, + { + "epoch": 4.974874371859297, + "grad_norm": 0.10158302634954453, + "learning_rate": 2.691538235974118e-05, + "loss": 0.0205, + "step": 2970 + }, + { + "epoch": 4.991624790619765, + "grad_norm": 0.10737495869398117, + "learning_rate": 2.6890227189923285e-05, + "loss": 0.0189, + "step": 2980 + }, + { + "epoch": 5.008375209380235, + "grad_norm": 0.07879546284675598, + "learning_rate": 2.6864981721733317e-05, + "loss": 0.0179, + "step": 2990 + }, + { + "epoch": 5.025125628140704, + "grad_norm": 0.10061459243297577, + "learning_rate": 2.683964614689383e-05, + "loss": 0.0196, + "step": 3000 + }, + { + "epoch": 5.041876046901173, + "grad_norm": 0.05866040289402008, + "learning_rate": 2.681422065781169e-05, + "loss": 0.0156, + "step": 3010 + }, + { + "epoch": 5.058626465661642, + "grad_norm": 0.13595989346504211, + "learning_rate": 2.6788705447576563e-05, + "loss": 0.0185, + "step": 3020 + }, + { + "epoch": 5.075376884422111, + "grad_norm": 0.10418418049812317, + "learning_rate": 2.6763100709959524e-05, + "loss": 0.0172, + "step": 3030 + }, + { + "epoch": 5.09212730318258, + "grad_norm": 0.09122584760189056, + "learning_rate": 2.6737406639411537e-05, + "loss": 0.0193, + "step": 3040 + }, + { + "epoch": 5.108877721943049, + "grad_norm": 0.10324976593255997, + "learning_rate": 2.6711623431061985e-05, + "loss": 0.0171, + "step": 3050 + }, + { + "epoch": 5.125628140703517, + "grad_norm": 0.11300911009311676, + "learning_rate": 2.6685751280717195e-05, + "loss": 0.0176, + "step": 3060 + }, + { + "epoch": 5.142378559463987, + "grad_norm": 0.08261276036500931, + "learning_rate": 2.665979038485896e-05, + "loss": 0.0195, + "step": 3070 + }, + { + "epoch": 5.159128978224456, + "grad_norm": 0.10486511141061783, + "learning_rate": 2.6633740940643022e-05, + "loss": 0.017, + "step": 3080 + }, + { + "epoch": 5.175879396984925, + "grad_norm": 0.10391178727149963, + "learning_rate": 2.6607603145897594e-05, + "loss": 0.0201, + "step": 3090 + }, + { + "epoch": 5.192629815745394, + "grad_norm": 0.0923885777592659, + "learning_rate": 2.6581377199121853e-05, + "loss": 0.0191, + "step": 3100 + }, + { + "epoch": 5.209380234505863, + "grad_norm": 0.12680231034755707, + "learning_rate": 2.6555063299484434e-05, + "loss": 0.0177, + "step": 3110 + }, + { + "epoch": 5.226130653266332, + "grad_norm": 0.07475026696920395, + "learning_rate": 2.6528661646821917e-05, + "loss": 0.0193, + "step": 3120 + }, + { + "epoch": 5.242881072026801, + "grad_norm": 0.10191482305526733, + "learning_rate": 2.6502172441637302e-05, + "loss": 0.0193, + "step": 3130 + }, + { + "epoch": 5.259631490787269, + "grad_norm": 0.08599104732275009, + "learning_rate": 2.6475595885098493e-05, + "loss": 0.0193, + "step": 3140 + }, + { + "epoch": 5.276381909547739, + "grad_norm": 0.0979737862944603, + "learning_rate": 2.644893217903677e-05, + "loss": 0.0177, + "step": 3150 + }, + { + "epoch": 5.293132328308207, + "grad_norm": 0.1444755345582962, + "learning_rate": 2.6422181525945265e-05, + "loss": 0.0185, + "step": 3160 + }, + { + "epoch": 5.309882747068677, + "grad_norm": 0.0918390303850174, + "learning_rate": 2.63953441289774e-05, + "loss": 0.02, + "step": 3170 + }, + { + "epoch": 5.326633165829146, + "grad_norm": 0.11050040274858475, + "learning_rate": 2.6368420191945365e-05, + "loss": 0.0181, + "step": 3180 + }, + { + "epoch": 5.343383584589615, + "grad_norm": 0.0900280773639679, + "learning_rate": 2.6341409919318572e-05, + "loss": 0.017, + "step": 3190 + }, + { + "epoch": 5.360134003350084, + "grad_norm": 0.08904273808002472, + "learning_rate": 2.6314313516222084e-05, + "loss": 0.0195, + "step": 3200 + }, + { + "epoch": 5.376884422110553, + "grad_norm": 0.10394366830587387, + "learning_rate": 2.6287131188435075e-05, + "loss": 0.0193, + "step": 3210 + }, + { + "epoch": 5.393634840871022, + "grad_norm": 0.10032975673675537, + "learning_rate": 2.6259863142389252e-05, + "loss": 0.0164, + "step": 3220 + }, + { + "epoch": 5.410385259631491, + "grad_norm": 0.09366985410451889, + "learning_rate": 2.62325095851673e-05, + "loss": 0.0171, + "step": 3230 + }, + { + "epoch": 5.42713567839196, + "grad_norm": 0.08237487822771072, + "learning_rate": 2.620507072450131e-05, + "loss": 0.0191, + "step": 3240 + }, + { + "epoch": 5.443886097152429, + "grad_norm": 0.10653271526098251, + "learning_rate": 2.6177546768771183e-05, + "loss": 0.0173, + "step": 3250 + }, + { + "epoch": 5.460636515912898, + "grad_norm": 0.08800845593214035, + "learning_rate": 2.6149937927003062e-05, + "loss": 0.018, + "step": 3260 + }, + { + "epoch": 5.477386934673367, + "grad_norm": 0.10618974268436432, + "learning_rate": 2.6122244408867766e-05, + "loss": 0.0198, + "step": 3270 + }, + { + "epoch": 5.494137353433836, + "grad_norm": 0.0838366374373436, + "learning_rate": 2.6094466424679148e-05, + "loss": 0.0196, + "step": 3280 + }, + { + "epoch": 5.510887772194305, + "grad_norm": 0.08228679746389389, + "learning_rate": 2.606660418539253e-05, + "loss": 0.0208, + "step": 3290 + }, + { + "epoch": 5.527638190954773, + "grad_norm": 0.10232897847890854, + "learning_rate": 2.6038657902603107e-05, + "loss": 0.0173, + "step": 3300 + }, + { + "epoch": 5.544388609715243, + "grad_norm": 0.0675511434674263, + "learning_rate": 2.6010627788544314e-05, + "loss": 0.0171, + "step": 3310 + }, + { + "epoch": 5.561139028475711, + "grad_norm": 0.11169396340847015, + "learning_rate": 2.5982514056086238e-05, + "loss": 0.0202, + "step": 3320 + }, + { + "epoch": 5.577889447236181, + "grad_norm": 0.09199520200490952, + "learning_rate": 2.5954316918733992e-05, + "loss": 0.0188, + "step": 3330 + }, + { + "epoch": 5.594639865996649, + "grad_norm": 0.09636520594358444, + "learning_rate": 2.5926036590626093e-05, + "loss": 0.0189, + "step": 3340 + }, + { + "epoch": 5.611390284757119, + "grad_norm": 0.1068258136510849, + "learning_rate": 2.589767328653283e-05, + "loss": 0.0185, + "step": 3350 + }, + { + "epoch": 5.628140703517588, + "grad_norm": 0.0738004669547081, + "learning_rate": 2.5869227221854646e-05, + "loss": 0.0153, + "step": 3360 + }, + { + "epoch": 5.644891122278057, + "grad_norm": 0.08421333134174347, + "learning_rate": 2.584069861262049e-05, + "loss": 0.0199, + "step": 3370 + }, + { + "epoch": 5.661641541038526, + "grad_norm": 0.1111551895737648, + "learning_rate": 2.58120876754862e-05, + "loss": 0.0147, + "step": 3380 + }, + { + "epoch": 5.678391959798995, + "grad_norm": 0.0835203304886818, + "learning_rate": 2.5783394627732815e-05, + "loss": 0.0168, + "step": 3390 + }, + { + "epoch": 5.695142378559464, + "grad_norm": 0.12761108577251434, + "learning_rate": 2.5754619687264957e-05, + "loss": 0.0189, + "step": 3400 + }, + { + "epoch": 5.711892797319933, + "grad_norm": 0.0886048823595047, + "learning_rate": 2.5725763072609178e-05, + "loss": 0.0164, + "step": 3410 + }, + { + "epoch": 5.728643216080402, + "grad_norm": 0.09086675196886063, + "learning_rate": 2.569682500291229e-05, + "loss": 0.0151, + "step": 3420 + }, + { + "epoch": 5.745393634840871, + "grad_norm": 0.10654979199171066, + "learning_rate": 2.5667805697939687e-05, + "loss": 0.0173, + "step": 3430 + }, + { + "epoch": 5.76214405360134, + "grad_norm": 0.09250108897686005, + "learning_rate": 2.5638705378073704e-05, + "loss": 0.0207, + "step": 3440 + }, + { + "epoch": 5.778894472361809, + "grad_norm": 0.09105037152767181, + "learning_rate": 2.5609524264311934e-05, + "loss": 0.0175, + "step": 3450 + }, + { + "epoch": 5.795644891122278, + "grad_norm": 0.10493557155132294, + "learning_rate": 2.5580262578265536e-05, + "loss": 0.0217, + "step": 3460 + }, + { + "epoch": 5.812395309882747, + "grad_norm": 0.09852892905473709, + "learning_rate": 2.5550920542157557e-05, + "loss": 0.0153, + "step": 3470 + }, + { + "epoch": 5.8291457286432165, + "grad_norm": 0.09802744537591934, + "learning_rate": 2.5521498378821273e-05, + "loss": 0.0195, + "step": 3480 + }, + { + "epoch": 5.8458961474036855, + "grad_norm": 0.07921155542135239, + "learning_rate": 2.549199631169844e-05, + "loss": 0.0171, + "step": 3490 + }, + { + "epoch": 5.8626465661641545, + "grad_norm": 0.09621422737836838, + "learning_rate": 2.5462414564837663e-05, + "loss": 0.0168, + "step": 3500 + }, + { + "epoch": 5.8793969849246235, + "grad_norm": 0.0989849641919136, + "learning_rate": 2.543275336289264e-05, + "loss": 0.017, + "step": 3510 + }, + { + "epoch": 5.8961474036850925, + "grad_norm": 0.11342565715312958, + "learning_rate": 2.540301293112048e-05, + "loss": 0.018, + "step": 3520 + }, + { + "epoch": 5.9128978224455615, + "grad_norm": 0.1399437040090561, + "learning_rate": 2.537319349538e-05, + "loss": 0.0182, + "step": 3530 + }, + { + "epoch": 5.9296482412060305, + "grad_norm": 0.09442800283432007, + "learning_rate": 2.534329528212999e-05, + "loss": 0.0164, + "step": 3540 + }, + { + "epoch": 5.9463986599664995, + "grad_norm": 0.09357256442308426, + "learning_rate": 2.531331851842751e-05, + "loss": 0.017, + "step": 3550 + }, + { + "epoch": 5.9631490787269685, + "grad_norm": 0.10102587193250656, + "learning_rate": 2.528326343192615e-05, + "loss": 0.0161, + "step": 3560 + }, + { + "epoch": 5.9798994974874375, + "grad_norm": 0.09273260831832886, + "learning_rate": 2.5253130250874316e-05, + "loss": 0.0168, + "step": 3570 + }, + { + "epoch": 5.9966499162479066, + "grad_norm": 0.09276672452688217, + "learning_rate": 2.5222919204113494e-05, + "loss": 0.0198, + "step": 3580 + }, + { + "epoch": 6.013400335008376, + "grad_norm": 0.15660251677036285, + "learning_rate": 2.5192630521076494e-05, + "loss": 0.0171, + "step": 3590 + }, + { + "epoch": 6.030150753768845, + "grad_norm": 0.10578886419534683, + "learning_rate": 2.5162264431785734e-05, + "loss": 0.0199, + "step": 3600 + }, + { + "epoch": 6.046901172529314, + "grad_norm": 0.10759575664997101, + "learning_rate": 2.5131821166851484e-05, + "loss": 0.0167, + "step": 3610 + }, + { + "epoch": 6.063651591289783, + "grad_norm": 0.08797171711921692, + "learning_rate": 2.5101300957470095e-05, + "loss": 0.0177, + "step": 3620 + }, + { + "epoch": 6.080402010050252, + "grad_norm": 0.06793040037155151, + "learning_rate": 2.507070403542228e-05, + "loss": 0.0166, + "step": 3630 + }, + { + "epoch": 6.097152428810721, + "grad_norm": 0.08354511857032776, + "learning_rate": 2.504003063307132e-05, + "loss": 0.0177, + "step": 3640 + }, + { + "epoch": 6.11390284757119, + "grad_norm": 0.06704747676849365, + "learning_rate": 2.500928098336131e-05, + "loss": 0.0184, + "step": 3650 + }, + { + "epoch": 6.130653266331659, + "grad_norm": 0.122153140604496, + "learning_rate": 2.497845531981541e-05, + "loss": 0.0191, + "step": 3660 + }, + { + "epoch": 6.147403685092128, + "grad_norm": 0.08629975467920303, + "learning_rate": 2.4947553876534047e-05, + "loss": 0.0175, + "step": 3670 + }, + { + "epoch": 6.164154103852597, + "grad_norm": 0.09296640753746033, + "learning_rate": 2.4916576888193137e-05, + "loss": 0.0148, + "step": 3680 + }, + { + "epoch": 6.180904522613066, + "grad_norm": 0.06364050507545471, + "learning_rate": 2.488552459004231e-05, + "loss": 0.0169, + "step": 3690 + }, + { + "epoch": 6.197654941373535, + "grad_norm": 0.07617541402578354, + "learning_rate": 2.4854397217903145e-05, + "loss": 0.0176, + "step": 3700 + }, + { + "epoch": 6.214405360134004, + "grad_norm": 0.08891832083463669, + "learning_rate": 2.4823195008167325e-05, + "loss": 0.0172, + "step": 3710 + }, + { + "epoch": 6.231155778894473, + "grad_norm": 0.06869089603424072, + "learning_rate": 2.4791918197794907e-05, + "loss": 0.0161, + "step": 3720 + }, + { + "epoch": 6.247906197654942, + "grad_norm": 0.10788177698850632, + "learning_rate": 2.4760567024312476e-05, + "loss": 0.0154, + "step": 3730 + }, + { + "epoch": 6.264656616415411, + "grad_norm": 0.14439617097377777, + "learning_rate": 2.4729141725811344e-05, + "loss": 0.0183, + "step": 3740 + }, + { + "epoch": 6.28140703517588, + "grad_norm": 0.10388529300689697, + "learning_rate": 2.4697642540945774e-05, + "loss": 0.0194, + "step": 3750 + }, + { + "epoch": 6.298157453936349, + "grad_norm": 0.08002973347902298, + "learning_rate": 2.4666069708931136e-05, + "loss": 0.015, + "step": 3760 + }, + { + "epoch": 6.314907872696818, + "grad_norm": 0.0883835032582283, + "learning_rate": 2.46344234695421e-05, + "loss": 0.0183, + "step": 3770 + }, + { + "epoch": 6.331658291457287, + "grad_norm": 0.08702576160430908, + "learning_rate": 2.460270406311082e-05, + "loss": 0.0159, + "step": 3780 + }, + { + "epoch": 6.348408710217756, + "grad_norm": 0.09502051025629044, + "learning_rate": 2.457091173052511e-05, + "loss": 0.0173, + "step": 3790 + }, + { + "epoch": 6.365159128978225, + "grad_norm": 0.09019004553556442, + "learning_rate": 2.4539046713226593e-05, + "loss": 0.0192, + "step": 3800 + }, + { + "epoch": 6.381909547738694, + "grad_norm": 0.10312219709157944, + "learning_rate": 2.4507109253208898e-05, + "loss": 0.0154, + "step": 3810 + }, + { + "epoch": 6.398659966499163, + "grad_norm": 0.08851608633995056, + "learning_rate": 2.4475099593015813e-05, + "loss": 0.0172, + "step": 3820 + }, + { + "epoch": 6.415410385259632, + "grad_norm": 0.0891028419137001, + "learning_rate": 2.4443017975739423e-05, + "loss": 0.0188, + "step": 3830 + }, + { + "epoch": 6.432160804020101, + "grad_norm": 0.08572867512702942, + "learning_rate": 2.4410864645018288e-05, + "loss": 0.0165, + "step": 3840 + }, + { + "epoch": 6.44891122278057, + "grad_norm": 0.09533847123384476, + "learning_rate": 2.4378639845035583e-05, + "loss": 0.0158, + "step": 3850 + }, + { + "epoch": 6.465661641541039, + "grad_norm": 0.0869334265589714, + "learning_rate": 2.4346343820517237e-05, + "loss": 0.0158, + "step": 3860 + }, + { + "epoch": 6.482412060301508, + "grad_norm": 0.10855986922979355, + "learning_rate": 2.4313976816730095e-05, + "loss": 0.0179, + "step": 3870 + }, + { + "epoch": 6.499162479061977, + "grad_norm": 0.0890968069434166, + "learning_rate": 2.4281539079480047e-05, + "loss": 0.0166, + "step": 3880 + }, + { + "epoch": 6.515912897822446, + "grad_norm": 0.09087546169757843, + "learning_rate": 2.4249030855110128e-05, + "loss": 0.0156, + "step": 3890 + }, + { + "epoch": 6.532663316582915, + "grad_norm": 0.07741452008485794, + "learning_rate": 2.42164523904987e-05, + "loss": 0.0163, + "step": 3900 + }, + { + "epoch": 6.549413735343384, + "grad_norm": 0.09288477897644043, + "learning_rate": 2.4183803933057554e-05, + "loss": 0.0165, + "step": 3910 + }, + { + "epoch": 6.566164154103853, + "grad_norm": 0.08580487966537476, + "learning_rate": 2.4151085730730013e-05, + "loss": 0.0201, + "step": 3920 + }, + { + "epoch": 6.582914572864322, + "grad_norm": 0.079682357609272, + "learning_rate": 2.4118298031989084e-05, + "loss": 0.0163, + "step": 3930 + }, + { + "epoch": 6.599664991624791, + "grad_norm": 0.07965271174907684, + "learning_rate": 2.4085441085835534e-05, + "loss": 0.0162, + "step": 3940 + }, + { + "epoch": 6.61641541038526, + "grad_norm": 0.10382688790559769, + "learning_rate": 2.405251514179604e-05, + "loss": 0.0172, + "step": 3950 + }, + { + "epoch": 6.633165829145729, + "grad_norm": 0.0887349396944046, + "learning_rate": 2.401952044992126e-05, + "loss": 0.0181, + "step": 3960 + }, + { + "epoch": 6.649916247906198, + "grad_norm": 0.10867097228765488, + "learning_rate": 2.3986457260783942e-05, + "loss": 0.016, + "step": 3970 + }, + { + "epoch": 6.666666666666667, + "grad_norm": 0.10828111320734024, + "learning_rate": 2.395332582547703e-05, + "loss": 0.0216, + "step": 3980 + }, + { + "epoch": 6.683417085427136, + "grad_norm": 0.08616909384727478, + "learning_rate": 2.3920126395611755e-05, + "loss": 0.019, + "step": 3990 + }, + { + "epoch": 6.700167504187605, + "grad_norm": 0.10397583991289139, + "learning_rate": 2.3886859223315725e-05, + "loss": 0.0199, + "step": 4000 + }, + { + "epoch": 6.716917922948074, + "grad_norm": 0.12109193205833435, + "learning_rate": 2.385352456123099e-05, + "loss": 0.0191, + "step": 4010 + }, + { + "epoch": 6.733668341708543, + "grad_norm": 0.07098208367824554, + "learning_rate": 2.3820122662512155e-05, + "loss": 0.0164, + "step": 4020 + }, + { + "epoch": 6.750418760469012, + "grad_norm": 0.1095917746424675, + "learning_rate": 2.3786653780824443e-05, + "loss": 0.0153, + "step": 4030 + }, + { + "epoch": 6.767169179229481, + "grad_norm": 0.08910483866930008, + "learning_rate": 2.3753118170341762e-05, + "loss": 0.0168, + "step": 4040 + }, + { + "epoch": 6.78391959798995, + "grad_norm": 0.07048118114471436, + "learning_rate": 2.3719516085744782e-05, + "loss": 0.0152, + "step": 4050 + }, + { + "epoch": 6.800670016750419, + "grad_norm": 0.1071673035621643, + "learning_rate": 2.3685847782218997e-05, + "loss": 0.0158, + "step": 4060 + }, + { + "epoch": 6.817420435510888, + "grad_norm": 0.10280796140432358, + "learning_rate": 2.3652113515452792e-05, + "loss": 0.0173, + "step": 4070 + }, + { + "epoch": 6.834170854271357, + "grad_norm": 0.10277356952428818, + "learning_rate": 2.3618313541635506e-05, + "loss": 0.0173, + "step": 4080 + }, + { + "epoch": 6.850921273031826, + "grad_norm": 0.09373802691698074, + "learning_rate": 2.358444811745547e-05, + "loss": 0.0166, + "step": 4090 + }, + { + "epoch": 6.867671691792295, + "grad_norm": 0.09374375641345978, + "learning_rate": 2.3550517500098068e-05, + "loss": 0.0171, + "step": 4100 + }, + { + "epoch": 6.884422110552764, + "grad_norm": 0.08589532971382141, + "learning_rate": 2.3516521947243787e-05, + "loss": 0.0177, + "step": 4110 + }, + { + "epoch": 6.901172529313233, + "grad_norm": 0.09759582579135895, + "learning_rate": 2.348246171706626e-05, + "loss": 0.0161, + "step": 4120 + }, + { + "epoch": 6.917922948073702, + "grad_norm": 0.08552340418100357, + "learning_rate": 2.3448337068230288e-05, + "loss": 0.0156, + "step": 4130 + }, + { + "epoch": 6.934673366834171, + "grad_norm": 0.07144278287887573, + "learning_rate": 2.3414148259889896e-05, + "loss": 0.0184, + "step": 4140 + }, + { + "epoch": 6.95142378559464, + "grad_norm": 0.08075377345085144, + "learning_rate": 2.3379895551686358e-05, + "loss": 0.0182, + "step": 4150 + }, + { + "epoch": 6.968174204355109, + "grad_norm": 0.09423919767141342, + "learning_rate": 2.334557920374623e-05, + "loss": 0.0173, + "step": 4160 + }, + { + "epoch": 6.984924623115578, + "grad_norm": 0.09013880789279938, + "learning_rate": 2.3311199476679367e-05, + "loss": 0.0172, + "step": 4170 + }, + { + "epoch": 7.001675041876047, + "grad_norm": 0.1004585400223732, + "learning_rate": 2.327675663157693e-05, + "loss": 0.0171, + "step": 4180 + }, + { + "epoch": 7.018425460636516, + "grad_norm": 0.09630786627531052, + "learning_rate": 2.3242250930009446e-05, + "loss": 0.0169, + "step": 4190 + }, + { + "epoch": 7.035175879396985, + "grad_norm": 0.09441886842250824, + "learning_rate": 2.3207682634024777e-05, + "loss": 0.016, + "step": 4200 + }, + { + "epoch": 7.051926298157454, + "grad_norm": 0.1110304594039917, + "learning_rate": 2.3173052006146155e-05, + "loss": 0.0163, + "step": 4210 + }, + { + "epoch": 7.068676716917923, + "grad_norm": 0.07521197199821472, + "learning_rate": 2.313835930937018e-05, + "loss": 0.0146, + "step": 4220 + }, + { + "epoch": 7.085427135678392, + "grad_norm": 0.06828752905130386, + "learning_rate": 2.310360480716483e-05, + "loss": 0.0156, + "step": 4230 + }, + { + "epoch": 7.102177554438861, + "grad_norm": 0.08863577991724014, + "learning_rate": 2.3068788763467434e-05, + "loss": 0.0154, + "step": 4240 + }, + { + "epoch": 7.11892797319933, + "grad_norm": 0.10143563151359558, + "learning_rate": 2.3033911442682708e-05, + "loss": 0.0171, + "step": 4250 + }, + { + "epoch": 7.135678391959799, + "grad_norm": 0.10039176791906357, + "learning_rate": 2.299897310968072e-05, + "loss": 0.0185, + "step": 4260 + }, + { + "epoch": 7.152428810720268, + "grad_norm": 0.06830760836601257, + "learning_rate": 2.2963974029794893e-05, + "loss": 0.0135, + "step": 4270 + }, + { + "epoch": 7.169179229480737, + "grad_norm": 0.06696446985006332, + "learning_rate": 2.2928914468819968e-05, + "loss": 0.016, + "step": 4280 + }, + { + "epoch": 7.185929648241206, + "grad_norm": 0.06527981907129288, + "learning_rate": 2.2893794693010002e-05, + "loss": 0.0149, + "step": 4290 + }, + { + "epoch": 7.202680067001675, + "grad_norm": 0.08876840025186539, + "learning_rate": 2.2858614969076358e-05, + "loss": 0.0179, + "step": 4300 + }, + { + "epoch": 7.219430485762144, + "grad_norm": 0.08040497452020645, + "learning_rate": 2.2823375564185645e-05, + "loss": 0.0172, + "step": 4310 + }, + { + "epoch": 7.236180904522613, + "grad_norm": 0.06119128316640854, + "learning_rate": 2.278807674595773e-05, + "loss": 0.0157, + "step": 4320 + }, + { + "epoch": 7.252931323283082, + "grad_norm": 0.07552856206893921, + "learning_rate": 2.275271878246367e-05, + "loss": 0.0161, + "step": 4330 + }, + { + "epoch": 7.269681742043551, + "grad_norm": 0.09174931794404984, + "learning_rate": 2.271730194222369e-05, + "loss": 0.0157, + "step": 4340 + }, + { + "epoch": 7.28643216080402, + "grad_norm": 0.10719872266054153, + "learning_rate": 2.2681826494205156e-05, + "loss": 0.0175, + "step": 4350 + }, + { + "epoch": 7.303182579564489, + "grad_norm": 0.08944938331842422, + "learning_rate": 2.2646292707820513e-05, + "loss": 0.0146, + "step": 4360 + }, + { + "epoch": 7.319932998324958, + "grad_norm": 0.08042101562023163, + "learning_rate": 2.2610700852925254e-05, + "loss": 0.0145, + "step": 4370 + }, + { + "epoch": 7.336683417085427, + "grad_norm": 0.10604419559240341, + "learning_rate": 2.2575051199815854e-05, + "loss": 0.0175, + "step": 4380 + }, + { + "epoch": 7.353433835845896, + "grad_norm": 0.07255196571350098, + "learning_rate": 2.253934401922773e-05, + "loss": 0.0161, + "step": 4390 + }, + { + "epoch": 7.370184254606365, + "grad_norm": 0.07271875441074371, + "learning_rate": 2.250357958233319e-05, + "loss": 0.0181, + "step": 4400 + }, + { + "epoch": 7.386934673366834, + "grad_norm": 0.08586633205413818, + "learning_rate": 2.2467758160739352e-05, + "loss": 0.0159, + "step": 4410 + }, + { + "epoch": 7.403685092127303, + "grad_norm": 0.07192263007164001, + "learning_rate": 2.2431880026486114e-05, + "loss": 0.0176, + "step": 4420 + }, + { + "epoch": 7.420435510887772, + "grad_norm": 0.0814385861158371, + "learning_rate": 2.2395945452044045e-05, + "loss": 0.0171, + "step": 4430 + }, + { + "epoch": 7.437185929648241, + "grad_norm": 0.0670303925871849, + "learning_rate": 2.2359954710312363e-05, + "loss": 0.0145, + "step": 4440 + }, + { + "epoch": 7.45393634840871, + "grad_norm": 0.08451966196298599, + "learning_rate": 2.232390807461682e-05, + "loss": 0.0148, + "step": 4450 + }, + { + "epoch": 7.4706867671691795, + "grad_norm": 0.09510994702577591, + "learning_rate": 2.2287805818707658e-05, + "loss": 0.016, + "step": 4460 + }, + { + "epoch": 7.4874371859296485, + "grad_norm": 0.08456354588270187, + "learning_rate": 2.2251648216757514e-05, + "loss": 0.014, + "step": 4470 + }, + { + "epoch": 7.5041876046901175, + "grad_norm": 0.11731875687837601, + "learning_rate": 2.2215435543359337e-05, + "loss": 0.0154, + "step": 4480 + }, + { + "epoch": 7.5209380234505865, + "grad_norm": 0.10068397969007492, + "learning_rate": 2.2179168073524308e-05, + "loss": 0.0161, + "step": 4490 + }, + { + "epoch": 7.5376884422110555, + "grad_norm": 0.09683054685592651, + "learning_rate": 2.2142846082679763e-05, + "loss": 0.0168, + "step": 4500 + }, + { + "epoch": 7.5544388609715245, + "grad_norm": 0.06540673226118088, + "learning_rate": 2.2106469846667078e-05, + "loss": 0.013, + "step": 4510 + }, + { + "epoch": 7.5711892797319935, + "grad_norm": 0.0816856399178505, + "learning_rate": 2.2070039641739578e-05, + "loss": 0.0154, + "step": 4520 + }, + { + "epoch": 7.5879396984924625, + "grad_norm": 0.07817541062831879, + "learning_rate": 2.2033555744560462e-05, + "loss": 0.017, + "step": 4530 + }, + { + "epoch": 7.6046901172529315, + "grad_norm": 0.08995489031076431, + "learning_rate": 2.1997018432200674e-05, + "loss": 0.0153, + "step": 4540 + }, + { + "epoch": 7.6214405360134005, + "grad_norm": 0.09150712937116623, + "learning_rate": 2.1960427982136818e-05, + "loss": 0.0154, + "step": 4550 + }, + { + "epoch": 7.63819095477387, + "grad_norm": 0.11141885071992874, + "learning_rate": 2.1923784672249038e-05, + "loss": 0.0159, + "step": 4560 + }, + { + "epoch": 7.654941373534339, + "grad_norm": 0.09261714667081833, + "learning_rate": 2.1887088780818918e-05, + "loss": 0.015, + "step": 4570 + }, + { + "epoch": 7.671691792294808, + "grad_norm": 0.11019844561815262, + "learning_rate": 2.1850340586527354e-05, + "loss": 0.0181, + "step": 4580 + }, + { + "epoch": 7.688442211055277, + "grad_norm": 0.09286614507436752, + "learning_rate": 2.1813540368452468e-05, + "loss": 0.0165, + "step": 4590 + }, + { + "epoch": 7.705192629815746, + "grad_norm": 0.07964478433132172, + "learning_rate": 2.1776688406067447e-05, + "loss": 0.0162, + "step": 4600 + }, + { + "epoch": 7.721943048576215, + "grad_norm": 0.09548459202051163, + "learning_rate": 2.1739784979238447e-05, + "loss": 0.0165, + "step": 4610 + }, + { + "epoch": 7.738693467336684, + "grad_norm": 0.06440986692905426, + "learning_rate": 2.1702830368222478e-05, + "loss": 0.0141, + "step": 4620 + }, + { + "epoch": 7.755443886097153, + "grad_norm": 0.07117027044296265, + "learning_rate": 2.166582485366523e-05, + "loss": 0.0147, + "step": 4630 + }, + { + "epoch": 7.772194304857622, + "grad_norm": 0.08931209146976471, + "learning_rate": 2.1628768716599e-05, + "loss": 0.0147, + "step": 4640 + }, + { + "epoch": 7.788944723618091, + "grad_norm": 0.09026764333248138, + "learning_rate": 2.1591662238440515e-05, + "loss": 0.0159, + "step": 4650 + }, + { + "epoch": 7.80569514237856, + "grad_norm": 0.1498933583498001, + "learning_rate": 2.1554505700988813e-05, + "loss": 0.0155, + "step": 4660 + }, + { + "epoch": 7.822445561139029, + "grad_norm": 0.06725804507732391, + "learning_rate": 2.1517299386423094e-05, + "loss": 0.016, + "step": 4670 + }, + { + "epoch": 7.839195979899498, + "grad_norm": 0.08815880864858627, + "learning_rate": 2.148004357730059e-05, + "loss": 0.0151, + "step": 4680 + }, + { + "epoch": 7.855946398659967, + "grad_norm": 0.06878040730953217, + "learning_rate": 2.1442738556554404e-05, + "loss": 0.0172, + "step": 4690 + }, + { + "epoch": 7.872696817420436, + "grad_norm": 0.06804270297288895, + "learning_rate": 2.1405384607491375e-05, + "loss": 0.0156, + "step": 4700 + }, + { + "epoch": 7.889447236180905, + "grad_norm": 0.070171058177948, + "learning_rate": 2.1367982013789913e-05, + "loss": 0.0153, + "step": 4710 + }, + { + "epoch": 7.906197654941374, + "grad_norm": 0.07167346775531769, + "learning_rate": 2.133053105949786e-05, + "loss": 0.0156, + "step": 4720 + }, + { + "epoch": 7.922948073701843, + "grad_norm": 0.09849503636360168, + "learning_rate": 2.129303202903032e-05, + "loss": 0.0165, + "step": 4730 + }, + { + "epoch": 7.939698492462312, + "grad_norm": 0.09628047794103622, + "learning_rate": 2.1255485207167504e-05, + "loss": 0.016, + "step": 4740 + }, + { + "epoch": 7.956448911222781, + "grad_norm": 0.07518996298313141, + "learning_rate": 2.1217890879052556e-05, + "loss": 0.0148, + "step": 4750 + }, + { + "epoch": 7.97319932998325, + "grad_norm": 0.09254742413759232, + "learning_rate": 2.1180249330189423e-05, + "loss": 0.0177, + "step": 4760 + }, + { + "epoch": 7.989949748743719, + "grad_norm": 0.10075143724679947, + "learning_rate": 2.1142560846440645e-05, + "loss": 0.0173, + "step": 4770 + }, + { + "epoch": 8.006700167504187, + "grad_norm": 0.0872223973274231, + "learning_rate": 2.1104825714025203e-05, + "loss": 0.0155, + "step": 4780 + }, + { + "epoch": 8.023450586264657, + "grad_norm": 0.07023284584283829, + "learning_rate": 2.106704421951634e-05, + "loss": 0.0169, + "step": 4790 + }, + { + "epoch": 8.040201005025125, + "grad_norm": 0.0736195370554924, + "learning_rate": 2.1029216649839404e-05, + "loss": 0.0153, + "step": 4800 + }, + { + "epoch": 8.056951423785595, + "grad_norm": 0.12984727323055267, + "learning_rate": 2.099134329226963e-05, + "loss": 0.0194, + "step": 4810 + }, + { + "epoch": 8.073701842546063, + "grad_norm": 0.06633501499891281, + "learning_rate": 2.0953424434430006e-05, + "loss": 0.0172, + "step": 4820 + }, + { + "epoch": 8.090452261306533, + "grad_norm": 0.07725346088409424, + "learning_rate": 2.091546036428905e-05, + "loss": 0.0167, + "step": 4830 + }, + { + "epoch": 8.107202680067001, + "grad_norm": 0.0776301696896553, + "learning_rate": 2.0877451370158633e-05, + "loss": 0.0167, + "step": 4840 + }, + { + "epoch": 8.12395309882747, + "grad_norm": 0.08715241402387619, + "learning_rate": 2.083939774069181e-05, + "loss": 0.0152, + "step": 4850 + }, + { + "epoch": 8.140703517587939, + "grad_norm": 0.05368698760867119, + "learning_rate": 2.080129976488059e-05, + "loss": 0.0158, + "step": 4860 + }, + { + "epoch": 8.157453936348409, + "grad_norm": 0.07227354496717453, + "learning_rate": 2.0763157732053795e-05, + "loss": 0.0167, + "step": 4870 + }, + { + "epoch": 8.174204355108877, + "grad_norm": 0.08462859690189362, + "learning_rate": 2.0724971931874796e-05, + "loss": 0.0155, + "step": 4880 + }, + { + "epoch": 8.190954773869347, + "grad_norm": 0.0951157882809639, + "learning_rate": 2.0686742654339363e-05, + "loss": 0.0151, + "step": 4890 + }, + { + "epoch": 8.207705192629815, + "grad_norm": 0.07463134825229645, + "learning_rate": 2.0648470189773455e-05, + "loss": 0.0136, + "step": 4900 + }, + { + "epoch": 8.224455611390285, + "grad_norm": 0.08849921822547913, + "learning_rate": 2.0610154828830997e-05, + "loss": 0.0145, + "step": 4910 + }, + { + "epoch": 8.241206030150753, + "grad_norm": 0.08144073933362961, + "learning_rate": 2.0571796862491686e-05, + "loss": 0.0173, + "step": 4920 + }, + { + "epoch": 8.257956448911223, + "grad_norm": 0.08935582637786865, + "learning_rate": 2.0533396582058783e-05, + "loss": 0.0173, + "step": 4930 + }, + { + "epoch": 8.274706867671691, + "grad_norm": 0.06497425585985184, + "learning_rate": 2.0494954279156893e-05, + "loss": 0.0159, + "step": 4940 + }, + { + "epoch": 8.291457286432161, + "grad_norm": 0.09040762484073639, + "learning_rate": 2.0456470245729748e-05, + "loss": 0.018, + "step": 4950 + }, + { + "epoch": 8.308207705192629, + "grad_norm": 0.11972896009683609, + "learning_rate": 2.0417944774038013e-05, + "loss": 0.016, + "step": 4960 + }, + { + "epoch": 8.324958123953099, + "grad_norm": 0.09560854732990265, + "learning_rate": 2.0379378156657026e-05, + "loss": 0.0151, + "step": 4970 + }, + { + "epoch": 8.341708542713567, + "grad_norm": 0.09106440097093582, + "learning_rate": 2.0340770686474618e-05, + "loss": 0.0169, + "step": 4980 + }, + { + "epoch": 8.358458961474037, + "grad_norm": 0.07660805433988571, + "learning_rate": 2.030212265668886e-05, + "loss": 0.0141, + "step": 4990 + }, + { + "epoch": 8.375209380234505, + "grad_norm": 0.08468059450387955, + "learning_rate": 2.0263434360805847e-05, + "loss": 0.0167, + "step": 5000 + }, + { + "epoch": 8.391959798994975, + "grad_norm": 0.0799565464258194, + "learning_rate": 2.0224706092637472e-05, + "loss": 0.0153, + "step": 5010 + }, + { + "epoch": 8.408710217755443, + "grad_norm": 0.10247821360826492, + "learning_rate": 2.018593814629919e-05, + "loss": 0.0168, + "step": 5020 + }, + { + "epoch": 8.425460636515913, + "grad_norm": 0.07618727535009384, + "learning_rate": 2.0147130816207786e-05, + "loss": 0.016, + "step": 5030 + }, + { + "epoch": 8.442211055276381, + "grad_norm": 0.11073528230190277, + "learning_rate": 2.0108284397079127e-05, + "loss": 0.015, + "step": 5040 + }, + { + "epoch": 8.458961474036851, + "grad_norm": 0.10041352361440659, + "learning_rate": 2.006939918392596e-05, + "loss": 0.0156, + "step": 5050 + }, + { + "epoch": 8.47571189279732, + "grad_norm": 0.06778070330619812, + "learning_rate": 2.0030475472055625e-05, + "loss": 0.0151, + "step": 5060 + }, + { + "epoch": 8.492462311557789, + "grad_norm": 0.07777587324380875, + "learning_rate": 1.9991513557067835e-05, + "loss": 0.0144, + "step": 5070 + }, + { + "epoch": 8.509212730318257, + "grad_norm": 0.06961089372634888, + "learning_rate": 1.9952513734852444e-05, + "loss": 0.0139, + "step": 5080 + }, + { + "epoch": 8.525963149078727, + "grad_norm": 0.07862704992294312, + "learning_rate": 1.991347630158717e-05, + "loss": 0.0148, + "step": 5090 + }, + { + "epoch": 8.542713567839195, + "grad_norm": 0.10534326732158661, + "learning_rate": 1.9874401553735385e-05, + "loss": 0.0161, + "step": 5100 + }, + { + "epoch": 8.559463986599665, + "grad_norm": 0.08768954873085022, + "learning_rate": 1.9835289788043814e-05, + "loss": 0.0137, + "step": 5110 + }, + { + "epoch": 8.576214405360133, + "grad_norm": 0.06274593621492386, + "learning_rate": 1.979614130154032e-05, + "loss": 0.0149, + "step": 5120 + }, + { + "epoch": 8.592964824120603, + "grad_norm": 0.0664563998579979, + "learning_rate": 1.9756956391531646e-05, + "loss": 0.013, + "step": 5130 + }, + { + "epoch": 8.609715242881071, + "grad_norm": 0.061676476150751114, + "learning_rate": 1.9717735355601126e-05, + "loss": 0.0162, + "step": 5140 + }, + { + "epoch": 8.626465661641541, + "grad_norm": 0.08172836154699326, + "learning_rate": 1.9678478491606462e-05, + "loss": 0.0174, + "step": 5150 + }, + { + "epoch": 8.64321608040201, + "grad_norm": 0.05784829333424568, + "learning_rate": 1.9639186097677442e-05, + "loss": 0.0149, + "step": 5160 + }, + { + "epoch": 8.65996649916248, + "grad_norm": 0.08542623370885849, + "learning_rate": 1.9599858472213682e-05, + "loss": 0.0147, + "step": 5170 + }, + { + "epoch": 8.676716917922947, + "grad_norm": 0.07957877963781357, + "learning_rate": 1.9560495913882353e-05, + "loss": 0.0133, + "step": 5180 + }, + { + "epoch": 8.693467336683417, + "grad_norm": 0.0810636356472969, + "learning_rate": 1.9521098721615923e-05, + "loss": 0.0145, + "step": 5190 + }, + { + "epoch": 8.710217755443885, + "grad_norm": 0.07732214033603668, + "learning_rate": 1.9481667194609882e-05, + "loss": 0.0145, + "step": 5200 + }, + { + "epoch": 8.726968174204355, + "grad_norm": 0.0780918151140213, + "learning_rate": 1.9442201632320464e-05, + "loss": 0.0147, + "step": 5210 + }, + { + "epoch": 8.743718592964823, + "grad_norm": 0.09253431856632233, + "learning_rate": 1.9402702334462383e-05, + "loss": 0.0155, + "step": 5220 + }, + { + "epoch": 8.760469011725293, + "grad_norm": 0.06418832391500473, + "learning_rate": 1.9363169601006552e-05, + "loss": 0.0126, + "step": 5230 + }, + { + "epoch": 8.777219430485761, + "grad_norm": 0.09778792411088943, + "learning_rate": 1.9323603732177798e-05, + "loss": 0.0146, + "step": 5240 + }, + { + "epoch": 8.793969849246231, + "grad_norm": 0.0891941487789154, + "learning_rate": 1.9284005028452607e-05, + "loss": 0.0141, + "step": 5250 + }, + { + "epoch": 8.8107202680067, + "grad_norm": 0.1019863709807396, + "learning_rate": 1.9244373790556803e-05, + "loss": 0.0154, + "step": 5260 + }, + { + "epoch": 8.82747068676717, + "grad_norm": 0.08988063782453537, + "learning_rate": 1.9204710319463297e-05, + "loss": 0.0151, + "step": 5270 + }, + { + "epoch": 8.844221105527637, + "grad_norm": 0.0668616071343422, + "learning_rate": 1.9165014916389792e-05, + "loss": 0.0128, + "step": 5280 + }, + { + "epoch": 8.860971524288107, + "grad_norm": 0.08318133652210236, + "learning_rate": 1.9125287882796483e-05, + "loss": 0.0159, + "step": 5290 + }, + { + "epoch": 8.877721943048575, + "grad_norm": 0.07382380962371826, + "learning_rate": 1.9085529520383783e-05, + "loss": 0.0142, + "step": 5300 + }, + { + "epoch": 8.894472361809045, + "grad_norm": 0.07555682957172394, + "learning_rate": 1.9045740131090037e-05, + "loss": 0.0142, + "step": 5310 + }, + { + "epoch": 8.911222780569513, + "grad_norm": 0.09456376731395721, + "learning_rate": 1.9005920017089196e-05, + "loss": 0.0139, + "step": 5320 + }, + { + "epoch": 8.927973199329983, + "grad_norm": 0.11305142194032669, + "learning_rate": 1.8966069480788562e-05, + "loss": 0.0146, + "step": 5330 + }, + { + "epoch": 8.944723618090451, + "grad_norm": 0.06259690970182419, + "learning_rate": 1.8926188824826477e-05, + "loss": 0.0154, + "step": 5340 + }, + { + "epoch": 8.961474036850921, + "grad_norm": 0.10811670124530792, + "learning_rate": 1.8886278352070002e-05, + "loss": 0.014, + "step": 5350 + }, + { + "epoch": 8.97822445561139, + "grad_norm": 0.06475009024143219, + "learning_rate": 1.884633836561265e-05, + "loss": 0.014, + "step": 5360 + }, + { + "epoch": 8.99497487437186, + "grad_norm": 0.08297079056501389, + "learning_rate": 1.8806369168772086e-05, + "loss": 0.0141, + "step": 5370 + }, + { + "epoch": 9.011725293132328, + "grad_norm": 0.079750195145607, + "learning_rate": 1.8766371065087778e-05, + "loss": 0.0142, + "step": 5380 + }, + { + "epoch": 9.028475711892797, + "grad_norm": 0.07751394808292389, + "learning_rate": 1.8726344358318736e-05, + "loss": 0.0155, + "step": 5390 + }, + { + "epoch": 9.045226130653266, + "grad_norm": 0.0738295242190361, + "learning_rate": 1.86862893524412e-05, + "loss": 0.0142, + "step": 5400 + }, + { + "epoch": 9.061976549413735, + "grad_norm": 0.11355958878993988, + "learning_rate": 1.864620635164631e-05, + "loss": 0.0161, + "step": 5410 + }, + { + "epoch": 9.078726968174204, + "grad_norm": 0.08234552294015884, + "learning_rate": 1.8606095660337813e-05, + "loss": 0.016, + "step": 5420 + }, + { + "epoch": 9.095477386934673, + "grad_norm": 0.09014493972063065, + "learning_rate": 1.8565957583129753e-05, + "loss": 0.0133, + "step": 5430 + }, + { + "epoch": 9.112227805695142, + "grad_norm": 0.08228293806314468, + "learning_rate": 1.852579242484414e-05, + "loss": 0.0152, + "step": 5440 + }, + { + "epoch": 9.128978224455611, + "grad_norm": 0.06658344715833664, + "learning_rate": 1.848560049050866e-05, + "loss": 0.0158, + "step": 5450 + }, + { + "epoch": 9.14572864321608, + "grad_norm": 0.08792341500520706, + "learning_rate": 1.8445382085354336e-05, + "loss": 0.0122, + "step": 5460 + }, + { + "epoch": 9.16247906197655, + "grad_norm": 0.07732446491718292, + "learning_rate": 1.8405137514813224e-05, + "loss": 0.0141, + "step": 5470 + }, + { + "epoch": 9.179229480737018, + "grad_norm": 0.06328639388084412, + "learning_rate": 1.8364867084516084e-05, + "loss": 0.0138, + "step": 5480 + }, + { + "epoch": 9.195979899497488, + "grad_norm": 0.05957376956939697, + "learning_rate": 1.8324571100290064e-05, + "loss": 0.0137, + "step": 5490 + }, + { + "epoch": 9.212730318257956, + "grad_norm": 0.0765262320637703, + "learning_rate": 1.8284249868156378e-05, + "loss": 0.0153, + "step": 5500 + }, + { + "epoch": 9.229480737018426, + "grad_norm": 0.0667581558227539, + "learning_rate": 1.8243903694327992e-05, + "loss": 0.0139, + "step": 5510 + }, + { + "epoch": 9.246231155778894, + "grad_norm": 0.07833325862884521, + "learning_rate": 1.8203532885207272e-05, + "loss": 0.0156, + "step": 5520 + }, + { + "epoch": 9.262981574539364, + "grad_norm": 0.06736069172620773, + "learning_rate": 1.8163137747383675e-05, + "loss": 0.0127, + "step": 5530 + }, + { + "epoch": 9.279731993299832, + "grad_norm": 0.0944928526878357, + "learning_rate": 1.812271858763143e-05, + "loss": 0.0144, + "step": 5540 + }, + { + "epoch": 9.296482412060302, + "grad_norm": 0.09697055071592331, + "learning_rate": 1.8082275712907187e-05, + "loss": 0.0151, + "step": 5550 + }, + { + "epoch": 9.31323283082077, + "grad_norm": 0.09629473835229874, + "learning_rate": 1.8041809430347688e-05, + "loss": 0.0176, + "step": 5560 + }, + { + "epoch": 9.32998324958124, + "grad_norm": 0.07615918666124344, + "learning_rate": 1.8001320047267465e-05, + "loss": 0.0148, + "step": 5570 + }, + { + "epoch": 9.346733668341708, + "grad_norm": 0.07980474084615707, + "learning_rate": 1.796080787115646e-05, + "loss": 0.0149, + "step": 5580 + }, + { + "epoch": 9.363484087102178, + "grad_norm": 0.10057661682367325, + "learning_rate": 1.7920273209677722e-05, + "loss": 0.0167, + "step": 5590 + }, + { + "epoch": 9.380234505862646, + "grad_norm": 0.0847344771027565, + "learning_rate": 1.7879716370665068e-05, + "loss": 0.016, + "step": 5600 + }, + { + "epoch": 9.396984924623116, + "grad_norm": 0.07230829447507858, + "learning_rate": 1.783913766212073e-05, + "loss": 0.0161, + "step": 5610 + }, + { + "epoch": 9.413735343383584, + "grad_norm": 0.0874628946185112, + "learning_rate": 1.779853739221302e-05, + "loss": 0.0161, + "step": 5620 + }, + { + "epoch": 9.430485762144054, + "grad_norm": 0.07002632319927216, + "learning_rate": 1.775791586927401e-05, + "loss": 0.0148, + "step": 5630 + }, + { + "epoch": 9.447236180904522, + "grad_norm": 0.09408432990312576, + "learning_rate": 1.771727340179716e-05, + "loss": 0.013, + "step": 5640 + }, + { + "epoch": 9.463986599664992, + "grad_norm": 0.07826568186283112, + "learning_rate": 1.7676610298434993e-05, + "loss": 0.0155, + "step": 5650 + }, + { + "epoch": 9.48073701842546, + "grad_norm": 0.09074067324399948, + "learning_rate": 1.7635926867996758e-05, + "loss": 0.0144, + "step": 5660 + }, + { + "epoch": 9.49748743718593, + "grad_norm": 0.10472352802753448, + "learning_rate": 1.7595223419446064e-05, + "loss": 0.0159, + "step": 5670 + }, + { + "epoch": 9.514237855946398, + "grad_norm": 0.05583181977272034, + "learning_rate": 1.7554500261898545e-05, + "loss": 0.0138, + "step": 5680 + }, + { + "epoch": 9.530988274706868, + "grad_norm": 0.07133681327104568, + "learning_rate": 1.7513757704619513e-05, + "loss": 0.016, + "step": 5690 + }, + { + "epoch": 9.547738693467338, + "grad_norm": 0.057500433176755905, + "learning_rate": 1.747299605702161e-05, + "loss": 0.0135, + "step": 5700 + }, + { + "epoch": 9.564489112227806, + "grad_norm": 0.10911339521408081, + "learning_rate": 1.7432215628662463e-05, + "loss": 0.014, + "step": 5710 + }, + { + "epoch": 9.581239530988274, + "grad_norm": 0.09471974521875381, + "learning_rate": 1.7391416729242307e-05, + "loss": 0.0142, + "step": 5720 + }, + { + "epoch": 9.597989949748744, + "grad_norm": 0.08386453986167908, + "learning_rate": 1.7350599668601666e-05, + "loss": 0.0152, + "step": 5730 + }, + { + "epoch": 9.614740368509214, + "grad_norm": 0.09431836754083633, + "learning_rate": 1.730976475671899e-05, + "loss": 0.0181, + "step": 5740 + }, + { + "epoch": 9.631490787269682, + "grad_norm": 0.07204046100378036, + "learning_rate": 1.726891230370829e-05, + "loss": 0.0138, + "step": 5750 + }, + { + "epoch": 9.64824120603015, + "grad_norm": 0.08191115409135818, + "learning_rate": 1.7228042619816786e-05, + "loss": 0.0132, + "step": 5760 + }, + { + "epoch": 9.66499162479062, + "grad_norm": 0.06609786301851273, + "learning_rate": 1.7187156015422573e-05, + "loss": 0.0146, + "step": 5770 + }, + { + "epoch": 9.68174204355109, + "grad_norm": 0.08089634031057358, + "learning_rate": 1.7146252801032225e-05, + "loss": 0.014, + "step": 5780 + }, + { + "epoch": 9.698492462311558, + "grad_norm": 0.06636945903301239, + "learning_rate": 1.7105333287278465e-05, + "loss": 0.0126, + "step": 5790 + }, + { + "epoch": 9.715242881072026, + "grad_norm": 0.08253764361143112, + "learning_rate": 1.706439778491781e-05, + "loss": 0.0146, + "step": 5800 + }, + { + "epoch": 9.731993299832496, + "grad_norm": 0.06800912320613861, + "learning_rate": 1.702344660482818e-05, + "loss": 0.015, + "step": 5810 + }, + { + "epoch": 9.748743718592966, + "grad_norm": 0.05859517678618431, + "learning_rate": 1.6982480058006575e-05, + "loss": 0.0166, + "step": 5820 + }, + { + "epoch": 9.765494137353434, + "grad_norm": 0.09688255935907364, + "learning_rate": 1.6941498455566684e-05, + "loss": 0.0134, + "step": 5830 + }, + { + "epoch": 9.782244556113902, + "grad_norm": 0.07653312385082245, + "learning_rate": 1.690050210873654e-05, + "loss": 0.014, + "step": 5840 + }, + { + "epoch": 9.798994974874372, + "grad_norm": 0.08134609460830688, + "learning_rate": 1.6859491328856143e-05, + "loss": 0.0133, + "step": 5850 + }, + { + "epoch": 9.815745393634842, + "grad_norm": 0.06555905193090439, + "learning_rate": 1.6818466427375114e-05, + "loss": 0.0149, + "step": 5860 + }, + { + "epoch": 9.83249581239531, + "grad_norm": 0.06328080594539642, + "learning_rate": 1.67774277158503e-05, + "loss": 0.0134, + "step": 5870 + }, + { + "epoch": 9.849246231155778, + "grad_norm": 0.08666656166315079, + "learning_rate": 1.673637550594344e-05, + "loss": 0.0146, + "step": 5880 + }, + { + "epoch": 9.865996649916248, + "grad_norm": 0.06649037450551987, + "learning_rate": 1.669531010941879e-05, + "loss": 0.013, + "step": 5890 + }, + { + "epoch": 9.882747068676718, + "grad_norm": 0.0830390453338623, + "learning_rate": 1.6654231838140723e-05, + "loss": 0.013, + "step": 5900 + }, + { + "epoch": 9.899497487437186, + "grad_norm": 0.07208152860403061, + "learning_rate": 1.661314100407142e-05, + "loss": 0.0141, + "step": 5910 + }, + { + "epoch": 9.916247906197654, + "grad_norm": 0.07338222116231918, + "learning_rate": 1.657203791926845e-05, + "loss": 0.0123, + "step": 5920 + }, + { + "epoch": 9.932998324958124, + "grad_norm": 0.09653204679489136, + "learning_rate": 1.653092289588241e-05, + "loss": 0.0144, + "step": 5930 + }, + { + "epoch": 9.949748743718594, + "grad_norm": 0.07927566021680832, + "learning_rate": 1.6489796246154586e-05, + "loss": 0.0143, + "step": 5940 + }, + { + "epoch": 9.966499162479062, + "grad_norm": 0.057749975472688675, + "learning_rate": 1.6448658282414543e-05, + "loss": 0.0139, + "step": 5950 + }, + { + "epoch": 9.98324958123953, + "grad_norm": 0.07733950763940811, + "learning_rate": 1.6407509317077764e-05, + "loss": 0.0145, + "step": 5960 + }, + { + "epoch": 10.0, + "grad_norm": 0.06586837023496628, + "learning_rate": 1.6366349662643297e-05, + "loss": 0.0135, + "step": 5970 + }, + { + "epoch": 10.01675041876047, + "grad_norm": 0.09854725748300552, + "learning_rate": 1.6325179631691355e-05, + "loss": 0.0145, + "step": 5980 + }, + { + "epoch": 10.033500837520938, + "grad_norm": 0.08377289772033691, + "learning_rate": 1.6283999536880957e-05, + "loss": 0.0136, + "step": 5990 + }, + { + "epoch": 10.050251256281408, + "grad_norm": 0.07842150330543518, + "learning_rate": 1.6242809690947555e-05, + "loss": 0.0123, + "step": 6000 + }, + { + "epoch": 10.067001675041876, + "grad_norm": 0.08640950173139572, + "learning_rate": 1.6201610406700645e-05, + "loss": 0.0133, + "step": 6010 + }, + { + "epoch": 10.083752093802346, + "grad_norm": 0.1089024543762207, + "learning_rate": 1.6160401997021406e-05, + "loss": 0.014, + "step": 6020 + }, + { + "epoch": 10.100502512562814, + "grad_norm": 0.06064023822546005, + "learning_rate": 1.6119184774860322e-05, + "loss": 0.0146, + "step": 6030 + }, + { + "epoch": 10.117252931323284, + "grad_norm": 0.06450966000556946, + "learning_rate": 1.6077959053234796e-05, + "loss": 0.013, + "step": 6040 + }, + { + "epoch": 10.134003350083752, + "grad_norm": 0.07623566687107086, + "learning_rate": 1.6036725145226774e-05, + "loss": 0.0138, + "step": 6050 + }, + { + "epoch": 10.150753768844222, + "grad_norm": 0.0815720483660698, + "learning_rate": 1.5995483363980394e-05, + "loss": 0.0138, + "step": 6060 + }, + { + "epoch": 10.16750418760469, + "grad_norm": 0.061299752444028854, + "learning_rate": 1.595423402269956e-05, + "loss": 0.0137, + "step": 6070 + }, + { + "epoch": 10.18425460636516, + "grad_norm": 0.05670687556266785, + "learning_rate": 1.5912977434645608e-05, + "loss": 0.0165, + "step": 6080 + }, + { + "epoch": 10.201005025125628, + "grad_norm": 0.07310479134321213, + "learning_rate": 1.587171391313489e-05, + "loss": 0.0135, + "step": 6090 + }, + { + "epoch": 10.217755443886098, + "grad_norm": 0.06537134945392609, + "learning_rate": 1.5830443771536434e-05, + "loss": 0.0144, + "step": 6100 + }, + { + "epoch": 10.234505862646566, + "grad_norm": 0.07251140475273132, + "learning_rate": 1.578916732326953e-05, + "loss": 0.0157, + "step": 6110 + }, + { + "epoch": 10.251256281407034, + "grad_norm": 0.09199927002191544, + "learning_rate": 1.5747884881801364e-05, + "loss": 0.0144, + "step": 6120 + }, + { + "epoch": 10.268006700167504, + "grad_norm": 0.06975611299276352, + "learning_rate": 1.570659676064464e-05, + "loss": 0.0141, + "step": 6130 + }, + { + "epoch": 10.284757118927974, + "grad_norm": 0.0650688111782074, + "learning_rate": 1.5665303273355197e-05, + "loss": 0.0139, + "step": 6140 + }, + { + "epoch": 10.301507537688442, + "grad_norm": 0.11455053091049194, + "learning_rate": 1.562400473352962e-05, + "loss": 0.0134, + "step": 6150 + }, + { + "epoch": 10.318257956448912, + "grad_norm": 0.10255176573991776, + "learning_rate": 1.5582701454802867e-05, + "loss": 0.0126, + "step": 6160 + }, + { + "epoch": 10.33500837520938, + "grad_norm": 0.09504007548093796, + "learning_rate": 1.5541393750845886e-05, + "loss": 0.013, + "step": 6170 + }, + { + "epoch": 10.35175879396985, + "grad_norm": 0.053999241441488266, + "learning_rate": 1.5500081935363233e-05, + "loss": 0.0128, + "step": 6180 + }, + { + "epoch": 10.368509212730318, + "grad_norm": 0.07180154323577881, + "learning_rate": 1.5458766322090684e-05, + "loss": 0.0131, + "step": 6190 + }, + { + "epoch": 10.385259631490788, + "grad_norm": 0.07966731488704681, + "learning_rate": 1.5417447224792862e-05, + "loss": 0.0144, + "step": 6200 + }, + { + "epoch": 10.402010050251256, + "grad_norm": 0.08139584958553314, + "learning_rate": 1.5376124957260847e-05, + "loss": 0.0151, + "step": 6210 + }, + { + "epoch": 10.418760469011726, + "grad_norm": 0.06023263931274414, + "learning_rate": 1.5334799833309792e-05, + "loss": 0.0131, + "step": 6220 + }, + { + "epoch": 10.435510887772194, + "grad_norm": 0.09527294337749481, + "learning_rate": 1.529347216677654e-05, + "loss": 0.0134, + "step": 6230 + }, + { + "epoch": 10.452261306532664, + "grad_norm": 0.07697620987892151, + "learning_rate": 1.5252142271517259e-05, + "loss": 0.0139, + "step": 6240 + }, + { + "epoch": 10.469011725293132, + "grad_norm": 0.08630616217851639, + "learning_rate": 1.5210810461405019e-05, + "loss": 0.0127, + "step": 6250 + }, + { + "epoch": 10.485762144053602, + "grad_norm": 0.08324746787548065, + "learning_rate": 1.5169477050327457e-05, + "loss": 0.0137, + "step": 6260 + }, + { + "epoch": 10.50251256281407, + "grad_norm": 0.10764531046152115, + "learning_rate": 1.5128142352184346e-05, + "loss": 0.0116, + "step": 6270 + }, + { + "epoch": 10.519262981574538, + "grad_norm": 0.09146895259618759, + "learning_rate": 1.5086806680885245e-05, + "loss": 0.0125, + "step": 6280 + }, + { + "epoch": 10.536013400335008, + "grad_norm": 0.06823863834142685, + "learning_rate": 1.5045470350347106e-05, + "loss": 0.0116, + "step": 6290 + }, + { + "epoch": 10.552763819095478, + "grad_norm": 0.08591805398464203, + "learning_rate": 1.5004133674491876e-05, + "loss": 0.0125, + "step": 6300 + }, + { + "epoch": 10.569514237855946, + "grad_norm": 0.09628190100193024, + "learning_rate": 1.4962796967244137e-05, + "loss": 0.014, + "step": 6310 + }, + { + "epoch": 10.586264656616414, + "grad_norm": 0.09630268812179565, + "learning_rate": 1.4921460542528706e-05, + "loss": 0.0141, + "step": 6320 + }, + { + "epoch": 10.603015075376884, + "grad_norm": 0.06456310302019119, + "learning_rate": 1.4880124714268248e-05, + "loss": 0.0129, + "step": 6330 + }, + { + "epoch": 10.619765494137354, + "grad_norm": 0.06395621597766876, + "learning_rate": 1.4838789796380901e-05, + "loss": 0.0137, + "step": 6340 + }, + { + "epoch": 10.636515912897822, + "grad_norm": 0.06975158303976059, + "learning_rate": 1.4797456102777899e-05, + "loss": 0.0159, + "step": 6350 + }, + { + "epoch": 10.653266331658292, + "grad_norm": 0.07676661759614944, + "learning_rate": 1.4756123947361164e-05, + "loss": 0.0129, + "step": 6360 + }, + { + "epoch": 10.67001675041876, + "grad_norm": 0.07465475797653198, + "learning_rate": 1.4714793644020942e-05, + "loss": 0.013, + "step": 6370 + }, + { + "epoch": 10.68676716917923, + "grad_norm": 0.08438120037317276, + "learning_rate": 1.4673465506633417e-05, + "loss": 0.0148, + "step": 6380 + }, + { + "epoch": 10.703517587939698, + "grad_norm": 0.0682619959115982, + "learning_rate": 1.4632139849058328e-05, + "loss": 0.0132, + "step": 6390 + }, + { + "epoch": 10.720268006700168, + "grad_norm": 0.05983544513583183, + "learning_rate": 1.4590816985136563e-05, + "loss": 0.0153, + "step": 6400 + }, + { + "epoch": 10.737018425460636, + "grad_norm": 0.06861865520477295, + "learning_rate": 1.4549497228687812e-05, + "loss": 0.0141, + "step": 6410 + }, + { + "epoch": 10.753768844221106, + "grad_norm": 0.07951993495225906, + "learning_rate": 1.4508180893508157e-05, + "loss": 0.0145, + "step": 6420 + }, + { + "epoch": 10.770519262981574, + "grad_norm": 0.09159310907125473, + "learning_rate": 1.4466868293367708e-05, + "loss": 0.0128, + "step": 6430 + }, + { + "epoch": 10.787269681742044, + "grad_norm": 0.06889096647500992, + "learning_rate": 1.4425559742008196e-05, + "loss": 0.0144, + "step": 6440 + }, + { + "epoch": 10.804020100502512, + "grad_norm": 0.07343833148479462, + "learning_rate": 1.438425555314061e-05, + "loss": 0.0133, + "step": 6450 + }, + { + "epoch": 10.820770519262982, + "grad_norm": 0.08436884731054306, + "learning_rate": 1.4342956040442816e-05, + "loss": 0.0128, + "step": 6460 + }, + { + "epoch": 10.83752093802345, + "grad_norm": 0.0685647502541542, + "learning_rate": 1.4301661517557155e-05, + "loss": 0.0134, + "step": 6470 + }, + { + "epoch": 10.85427135678392, + "grad_norm": 0.07545416057109833, + "learning_rate": 1.426037229808808e-05, + "loss": 0.0125, + "step": 6480 + }, + { + "epoch": 10.871021775544389, + "grad_norm": 0.06606476753950119, + "learning_rate": 1.4219088695599773e-05, + "loss": 0.0131, + "step": 6490 + }, + { + "epoch": 10.887772194304858, + "grad_norm": 0.06588952988386154, + "learning_rate": 1.4177811023613749e-05, + "loss": 0.012, + "step": 6500 + }, + { + "epoch": 10.904522613065327, + "grad_norm": 0.0672588050365448, + "learning_rate": 1.4136539595606493e-05, + "loss": 0.0141, + "step": 6510 + }, + { + "epoch": 10.921273031825796, + "grad_norm": 0.06932391226291656, + "learning_rate": 1.4095274725007064e-05, + "loss": 0.0135, + "step": 6520 + }, + { + "epoch": 10.938023450586265, + "grad_norm": 0.055177588015794754, + "learning_rate": 1.4054016725194727e-05, + "loss": 0.0118, + "step": 6530 + }, + { + "epoch": 10.954773869346734, + "grad_norm": 0.06075577810406685, + "learning_rate": 1.4012765909496567e-05, + "loss": 0.0118, + "step": 6540 + }, + { + "epoch": 10.971524288107203, + "grad_norm": 0.08867711573839188, + "learning_rate": 1.3971522591185111e-05, + "loss": 0.0141, + "step": 6550 + }, + { + "epoch": 10.988274706867672, + "grad_norm": 0.06817186623811722, + "learning_rate": 1.3930287083475943e-05, + "loss": 0.0121, + "step": 6560 + }, + { + "epoch": 11.00502512562814, + "grad_norm": 0.06277189403772354, + "learning_rate": 1.3889059699525338e-05, + "loss": 0.0142, + "step": 6570 + }, + { + "epoch": 11.02177554438861, + "grad_norm": 0.05601593852043152, + "learning_rate": 1.3847840752427874e-05, + "loss": 0.0125, + "step": 6580 + }, + { + "epoch": 11.038525963149079, + "grad_norm": 0.14753024280071259, + "learning_rate": 1.380663055521405e-05, + "loss": 0.0133, + "step": 6590 + }, + { + "epoch": 11.055276381909549, + "grad_norm": 0.07654228061437607, + "learning_rate": 1.3765429420847928e-05, + "loss": 0.0126, + "step": 6600 + }, + { + "epoch": 11.072026800670017, + "grad_norm": 0.08025944232940674, + "learning_rate": 1.3724237662224738e-05, + "loss": 0.0146, + "step": 6610 + }, + { + "epoch": 11.088777219430487, + "grad_norm": 0.06721494346857071, + "learning_rate": 1.3683055592168497e-05, + "loss": 0.0159, + "step": 6620 + }, + { + "epoch": 11.105527638190955, + "grad_norm": 0.058076489716768265, + "learning_rate": 1.3641883523429665e-05, + "loss": 0.0145, + "step": 6630 + }, + { + "epoch": 11.122278056951425, + "grad_norm": 0.10511145740747452, + "learning_rate": 1.3600721768682729e-05, + "loss": 0.016, + "step": 6640 + }, + { + "epoch": 11.139028475711893, + "grad_norm": 0.07932860404253006, + "learning_rate": 1.3559570640523858e-05, + "loss": 0.0148, + "step": 6650 + }, + { + "epoch": 11.155778894472363, + "grad_norm": 0.1065361350774765, + "learning_rate": 1.351843045146852e-05, + "loss": 0.013, + "step": 6660 + }, + { + "epoch": 11.17252931323283, + "grad_norm": 0.09081228077411652, + "learning_rate": 1.3477301513949103e-05, + "loss": 0.0141, + "step": 6670 + }, + { + "epoch": 11.1892797319933, + "grad_norm": 0.06857980042695999, + "learning_rate": 1.3436184140312546e-05, + "loss": 0.0114, + "step": 6680 + }, + { + "epoch": 11.206030150753769, + "grad_norm": 0.06644202023744583, + "learning_rate": 1.3395078642817976e-05, + "loss": 0.0115, + "step": 6690 + }, + { + "epoch": 11.222780569514239, + "grad_norm": 0.07019989937543869, + "learning_rate": 1.3353985333634321e-05, + "loss": 0.0134, + "step": 6700 + }, + { + "epoch": 11.239530988274707, + "grad_norm": 0.06523793935775757, + "learning_rate": 1.3312904524837948e-05, + "loss": 0.0135, + "step": 6710 + }, + { + "epoch": 11.256281407035177, + "grad_norm": 0.08410218358039856, + "learning_rate": 1.3271836528410299e-05, + "loss": 0.0147, + "step": 6720 + }, + { + "epoch": 11.273031825795645, + "grad_norm": 0.10058104246854782, + "learning_rate": 1.3230781656235505e-05, + "loss": 0.015, + "step": 6730 + }, + { + "epoch": 11.289782244556115, + "grad_norm": 0.053105153143405914, + "learning_rate": 1.3189740220098026e-05, + "loss": 0.0118, + "step": 6740 + }, + { + "epoch": 11.306532663316583, + "grad_norm": 0.06775388866662979, + "learning_rate": 1.31487125316803e-05, + "loss": 0.0161, + "step": 6750 + }, + { + "epoch": 11.323283082077053, + "grad_norm": 0.06501705944538116, + "learning_rate": 1.3107698902560344e-05, + "loss": 0.0153, + "step": 6760 + }, + { + "epoch": 11.34003350083752, + "grad_norm": 0.08275182545185089, + "learning_rate": 1.3066699644209404e-05, + "loss": 0.0116, + "step": 6770 + }, + { + "epoch": 11.35678391959799, + "grad_norm": 0.06643733382225037, + "learning_rate": 1.3025715067989606e-05, + "loss": 0.0155, + "step": 6780 + }, + { + "epoch": 11.373534338358459, + "grad_norm": 0.08800774067640305, + "learning_rate": 1.2984745485151559e-05, + "loss": 0.0141, + "step": 6790 + }, + { + "epoch": 11.390284757118929, + "grad_norm": 0.10282844305038452, + "learning_rate": 1.2943791206832008e-05, + "loss": 0.0144, + "step": 6800 + }, + { + "epoch": 11.407035175879397, + "grad_norm": 0.08494739234447479, + "learning_rate": 1.290285254405148e-05, + "loss": 0.0138, + "step": 6810 + }, + { + "epoch": 11.423785594639867, + "grad_norm": 0.07839672267436981, + "learning_rate": 1.2861929807711901e-05, + "loss": 0.0148, + "step": 6820 + }, + { + "epoch": 11.440536013400335, + "grad_norm": 0.06659016013145447, + "learning_rate": 1.2821023308594256e-05, + "loss": 0.0144, + "step": 6830 + }, + { + "epoch": 11.457286432160805, + "grad_norm": 0.10885549336671829, + "learning_rate": 1.2780133357356218e-05, + "loss": 0.013, + "step": 6840 + }, + { + "epoch": 11.474036850921273, + "grad_norm": 0.07157129049301147, + "learning_rate": 1.273926026452978e-05, + "loss": 0.0138, + "step": 6850 + }, + { + "epoch": 11.490787269681743, + "grad_norm": 0.05953820422291756, + "learning_rate": 1.2698404340518922e-05, + "loss": 0.0129, + "step": 6860 + }, + { + "epoch": 11.507537688442211, + "grad_norm": 0.03799299895763397, + "learning_rate": 1.2657565895597222e-05, + "loss": 0.0106, + "step": 6870 + }, + { + "epoch": 11.52428810720268, + "grad_norm": 0.0774364024400711, + "learning_rate": 1.2616745239905527e-05, + "loss": 0.0138, + "step": 6880 + }, + { + "epoch": 11.541038525963149, + "grad_norm": 0.07412974536418915, + "learning_rate": 1.2575942683449585e-05, + "loss": 0.0134, + "step": 6890 + }, + { + "epoch": 11.557788944723619, + "grad_norm": 0.07800350338220596, + "learning_rate": 1.2535158536097691e-05, + "loss": 0.014, + "step": 6900 + }, + { + "epoch": 11.574539363484087, + "grad_norm": 0.08868993818759918, + "learning_rate": 1.249439310757833e-05, + "loss": 0.0139, + "step": 6910 + }, + { + "epoch": 11.591289782244557, + "grad_norm": 0.07635314017534256, + "learning_rate": 1.2453646707477843e-05, + "loss": 0.0134, + "step": 6920 + }, + { + "epoch": 11.608040201005025, + "grad_norm": 0.08172111213207245, + "learning_rate": 1.2412919645238049e-05, + "loss": 0.0124, + "step": 6930 + }, + { + "epoch": 11.624790619765495, + "grad_norm": 0.06785856932401657, + "learning_rate": 1.2372212230153911e-05, + "loss": 0.0118, + "step": 6940 + }, + { + "epoch": 11.641541038525963, + "grad_norm": 0.07923078536987305, + "learning_rate": 1.2331524771371197e-05, + "loss": 0.0131, + "step": 6950 + }, + { + "epoch": 11.658291457286433, + "grad_norm": 0.05836864933371544, + "learning_rate": 1.2290857577884109e-05, + "loss": 0.0126, + "step": 6960 + }, + { + "epoch": 11.675041876046901, + "grad_norm": 0.07780711352825165, + "learning_rate": 1.2250210958532943e-05, + "loss": 0.0141, + "step": 6970 + }, + { + "epoch": 11.691792294807371, + "grad_norm": 0.08186667412519455, + "learning_rate": 1.2209585222001762e-05, + "loss": 0.0129, + "step": 6980 + }, + { + "epoch": 11.708542713567839, + "grad_norm": 0.08738984167575836, + "learning_rate": 1.216898067681603e-05, + "loss": 0.0129, + "step": 6990 + }, + { + "epoch": 11.725293132328309, + "grad_norm": 0.07625941932201385, + "learning_rate": 1.2128397631340274e-05, + "loss": 0.0127, + "step": 7000 + }, + { + "epoch": 11.742043551088777, + "grad_norm": 0.06589437276124954, + "learning_rate": 1.2087836393775756e-05, + "loss": 0.0126, + "step": 7010 + }, + { + "epoch": 11.758793969849247, + "grad_norm": 0.06688161194324493, + "learning_rate": 1.2047297272158112e-05, + "loss": 0.0129, + "step": 7020 + }, + { + "epoch": 11.775544388609715, + "grad_norm": 0.08182154595851898, + "learning_rate": 1.2006780574355017e-05, + "loss": 0.0127, + "step": 7030 + }, + { + "epoch": 11.792294807370185, + "grad_norm": 0.06617710739374161, + "learning_rate": 1.1966286608063866e-05, + "loss": 0.0145, + "step": 7040 + }, + { + "epoch": 11.809045226130653, + "grad_norm": 0.060059934854507446, + "learning_rate": 1.1925815680809419e-05, + "loss": 0.0114, + "step": 7050 + }, + { + "epoch": 11.825795644891123, + "grad_norm": 0.08330146223306656, + "learning_rate": 1.1885368099941459e-05, + "loss": 0.0127, + "step": 7060 + }, + { + "epoch": 11.842546063651591, + "grad_norm": 0.08132729679346085, + "learning_rate": 1.1844944172632485e-05, + "loss": 0.0133, + "step": 7070 + }, + { + "epoch": 11.859296482412061, + "grad_norm": 0.08308562636375427, + "learning_rate": 1.1804544205875353e-05, + "loss": 0.0136, + "step": 7080 + }, + { + "epoch": 11.87604690117253, + "grad_norm": 0.07975098490715027, + "learning_rate": 1.1764168506480953e-05, + "loss": 0.0132, + "step": 7090 + }, + { + "epoch": 11.892797319932999, + "grad_norm": 0.07059597223997116, + "learning_rate": 1.1723817381075889e-05, + "loss": 0.0126, + "step": 7100 + }, + { + "epoch": 11.909547738693467, + "grad_norm": 0.07069030404090881, + "learning_rate": 1.1683491136100135e-05, + "loss": 0.0127, + "step": 7110 + }, + { + "epoch": 11.926298157453937, + "grad_norm": 0.06913162767887115, + "learning_rate": 1.1643190077804713e-05, + "loss": 0.0124, + "step": 7120 + }, + { + "epoch": 11.943048576214405, + "grad_norm": 0.07565716654062271, + "learning_rate": 1.1602914512249379e-05, + "loss": 0.0139, + "step": 7130 + }, + { + "epoch": 11.959798994974875, + "grad_norm": 0.05990506708621979, + "learning_rate": 1.156266474530028e-05, + "loss": 0.0111, + "step": 7140 + }, + { + "epoch": 11.976549413735343, + "grad_norm": 0.0784129723906517, + "learning_rate": 1.152244108262764e-05, + "loss": 0.0099, + "step": 7150 + }, + { + "epoch": 11.993299832495813, + "grad_norm": 0.0859949141740799, + "learning_rate": 1.1482243829703444e-05, + "loss": 0.0129, + "step": 7160 + }, + { + "epoch": 12.010050251256281, + "grad_norm": 0.0738619789481163, + "learning_rate": 1.1442073291799106e-05, + "loss": 0.0122, + "step": 7170 + }, + { + "epoch": 12.026800670016751, + "grad_norm": 0.09288010001182556, + "learning_rate": 1.1401929773983155e-05, + "loss": 0.0134, + "step": 7180 + }, + { + "epoch": 12.04355108877722, + "grad_norm": 0.07321544736623764, + "learning_rate": 1.1361813581118936e-05, + "loss": 0.0145, + "step": 7190 + }, + { + "epoch": 12.06030150753769, + "grad_norm": 0.07173691689968109, + "learning_rate": 1.1321725017862268e-05, + "loss": 0.0115, + "step": 7200 + }, + { + "epoch": 12.077051926298157, + "grad_norm": 0.058314502239227295, + "learning_rate": 1.1281664388659134e-05, + "loss": 0.0128, + "step": 7210 + }, + { + "epoch": 12.093802345058627, + "grad_norm": 0.07083640247583389, + "learning_rate": 1.1241631997743384e-05, + "loss": 0.0101, + "step": 7220 + }, + { + "epoch": 12.110552763819095, + "grad_norm": 0.065085269510746, + "learning_rate": 1.1201628149134422e-05, + "loss": 0.0137, + "step": 7230 + }, + { + "epoch": 12.127303182579565, + "grad_norm": 0.0781814306974411, + "learning_rate": 1.116165314663489e-05, + "loss": 0.0146, + "step": 7240 + }, + { + "epoch": 12.144053601340033, + "grad_norm": 0.09068423509597778, + "learning_rate": 1.1121707293828353e-05, + "loss": 0.014, + "step": 7250 + }, + { + "epoch": 12.160804020100503, + "grad_norm": 0.08548370748758316, + "learning_rate": 1.1081790894077022e-05, + "loss": 0.0121, + "step": 7260 + }, + { + "epoch": 12.177554438860971, + "grad_norm": 0.07648133486509323, + "learning_rate": 1.104190425051941e-05, + "loss": 0.0131, + "step": 7270 + }, + { + "epoch": 12.194304857621441, + "grad_norm": 0.09778308868408203, + "learning_rate": 1.1002047666068066e-05, + "loss": 0.0139, + "step": 7280 + }, + { + "epoch": 12.21105527638191, + "grad_norm": 0.07223774492740631, + "learning_rate": 1.0962221443407257e-05, + "loss": 0.0137, + "step": 7290 + }, + { + "epoch": 12.22780569514238, + "grad_norm": 0.0755016952753067, + "learning_rate": 1.0922425884990673e-05, + "loss": 0.0119, + "step": 7300 + }, + { + "epoch": 12.244556113902847, + "grad_norm": 0.0679793432354927, + "learning_rate": 1.0882661293039124e-05, + "loss": 0.0127, + "step": 7310 + }, + { + "epoch": 12.261306532663317, + "grad_norm": 0.08829930424690247, + "learning_rate": 1.0842927969538258e-05, + "loss": 0.0141, + "step": 7320 + }, + { + "epoch": 12.278056951423785, + "grad_norm": 0.06518194079399109, + "learning_rate": 1.0803226216236256e-05, + "loss": 0.0125, + "step": 7330 + }, + { + "epoch": 12.294807370184255, + "grad_norm": 0.08196015655994415, + "learning_rate": 1.076355633464154e-05, + "loss": 0.0114, + "step": 7340 + }, + { + "epoch": 12.311557788944723, + "grad_norm": 0.06805971264839172, + "learning_rate": 1.0723918626020498e-05, + "loss": 0.0118, + "step": 7350 + }, + { + "epoch": 12.328308207705193, + "grad_norm": 0.0839708149433136, + "learning_rate": 1.0684313391395176e-05, + "loss": 0.0148, + "step": 7360 + }, + { + "epoch": 12.345058626465661, + "grad_norm": 0.07113301753997803, + "learning_rate": 1.0644740931541005e-05, + "loss": 0.015, + "step": 7370 + }, + { + "epoch": 12.361809045226131, + "grad_norm": 0.08832506835460663, + "learning_rate": 1.0605201546984517e-05, + "loss": 0.0148, + "step": 7380 + }, + { + "epoch": 12.3785594639866, + "grad_norm": 0.09089552611112595, + "learning_rate": 1.0565695538001056e-05, + "loss": 0.016, + "step": 7390 + }, + { + "epoch": 12.39530988274707, + "grad_norm": 0.06780517846345901, + "learning_rate": 1.0526223204612495e-05, + "loss": 0.0119, + "step": 7400 + }, + { + "epoch": 12.412060301507537, + "grad_norm": 0.07529979944229126, + "learning_rate": 1.0486784846584974e-05, + "loss": 0.0111, + "step": 7410 + }, + { + "epoch": 12.428810720268007, + "grad_norm": 0.08596140146255493, + "learning_rate": 1.0447380763426607e-05, + "loss": 0.0122, + "step": 7420 + }, + { + "epoch": 12.445561139028475, + "grad_norm": 0.06264451146125793, + "learning_rate": 1.040801125438521e-05, + "loss": 0.013, + "step": 7430 + }, + { + "epoch": 12.462311557788945, + "grad_norm": 0.06601212918758392, + "learning_rate": 1.0368676618446042e-05, + "loss": 0.0132, + "step": 7440 + }, + { + "epoch": 12.479061976549414, + "grad_norm": 0.08287645131349564, + "learning_rate": 1.0329377154329512e-05, + "loss": 0.0127, + "step": 7450 + }, + { + "epoch": 12.495812395309883, + "grad_norm": 0.10921191424131393, + "learning_rate": 1.0290113160488925e-05, + "loss": 0.0125, + "step": 7460 + }, + { + "epoch": 12.512562814070352, + "grad_norm": 0.07738841325044632, + "learning_rate": 1.0250884935108223e-05, + "loss": 0.0142, + "step": 7470 + }, + { + "epoch": 12.529313232830821, + "grad_norm": 0.07163051515817642, + "learning_rate": 1.0211692776099698e-05, + "loss": 0.0118, + "step": 7480 + }, + { + "epoch": 12.54606365159129, + "grad_norm": 0.07590950280427933, + "learning_rate": 1.0172536981101747e-05, + "loss": 0.0093, + "step": 7490 + }, + { + "epoch": 12.56281407035176, + "grad_norm": 0.08660171180963516, + "learning_rate": 1.0133417847476606e-05, + "loss": 0.012, + "step": 7500 + }, + { + "epoch": 12.579564489112228, + "grad_norm": 0.07874178141355515, + "learning_rate": 1.0094335672308095e-05, + "loss": 0.0117, + "step": 7510 + }, + { + "epoch": 12.596314907872697, + "grad_norm": 0.07014073431491852, + "learning_rate": 1.0055290752399353e-05, + "loss": 0.0138, + "step": 7520 + }, + { + "epoch": 12.613065326633166, + "grad_norm": 0.07678911834955215, + "learning_rate": 1.0016283384270594e-05, + "loss": 0.0117, + "step": 7530 + }, + { + "epoch": 12.629815745393635, + "grad_norm": 0.06955492496490479, + "learning_rate": 9.977313864156854e-06, + "loss": 0.0108, + "step": 7540 + }, + { + "epoch": 12.646566164154104, + "grad_norm": 0.08744408190250397, + "learning_rate": 9.938382488005729e-06, + "loss": 0.0116, + "step": 7550 + }, + { + "epoch": 12.663316582914574, + "grad_norm": 0.08456790447235107, + "learning_rate": 9.89948955147515e-06, + "loss": 0.0142, + "step": 7560 + }, + { + "epoch": 12.680067001675042, + "grad_norm": 0.08989357948303223, + "learning_rate": 9.860635349931116e-06, + "loss": 0.011, + "step": 7570 + }, + { + "epoch": 12.696817420435512, + "grad_norm": 0.05640851706266403, + "learning_rate": 9.821820178445455e-06, + "loss": 0.0116, + "step": 7580 + }, + { + "epoch": 12.71356783919598, + "grad_norm": 0.07872806489467621, + "learning_rate": 9.783044331793605e-06, + "loss": 0.0106, + "step": 7590 + }, + { + "epoch": 12.73031825795645, + "grad_norm": 0.08626449853181839, + "learning_rate": 9.744308104452338e-06, + "loss": 0.0118, + "step": 7600 + }, + { + "epoch": 12.747068676716918, + "grad_norm": 0.05731945484876633, + "learning_rate": 9.705611790597562e-06, + "loss": 0.0114, + "step": 7610 + }, + { + "epoch": 12.763819095477388, + "grad_norm": 0.06404105573892593, + "learning_rate": 9.666955684102043e-06, + "loss": 0.0123, + "step": 7620 + }, + { + "epoch": 12.780569514237856, + "grad_norm": 0.0529644712805748, + "learning_rate": 9.62834007853323e-06, + "loss": 0.0127, + "step": 7630 + }, + { + "epoch": 12.797319932998326, + "grad_norm": 0.10183175653219223, + "learning_rate": 9.589765267150975e-06, + "loss": 0.0123, + "step": 7640 + }, + { + "epoch": 12.814070351758794, + "grad_norm": 0.061799515038728714, + "learning_rate": 9.551231542905326e-06, + "loss": 0.0134, + "step": 7650 + }, + { + "epoch": 12.830820770519264, + "grad_norm": 0.04345851019024849, + "learning_rate": 9.512739198434312e-06, + "loss": 0.0137, + "step": 7660 + }, + { + "epoch": 12.847571189279732, + "grad_norm": 0.057996947318315506, + "learning_rate": 9.474288526061698e-06, + "loss": 0.0144, + "step": 7670 + }, + { + "epoch": 12.864321608040202, + "grad_norm": 0.07847338914871216, + "learning_rate": 9.435879817794784e-06, + "loss": 0.0123, + "step": 7680 + }, + { + "epoch": 12.88107202680067, + "grad_norm": 0.0713241770863533, + "learning_rate": 9.39751336532218e-06, + "loss": 0.0101, + "step": 7690 + }, + { + "epoch": 12.89782244556114, + "grad_norm": 0.07546516507863998, + "learning_rate": 9.359189460011591e-06, + "loss": 0.0139, + "step": 7700 + }, + { + "epoch": 12.914572864321608, + "grad_norm": 0.04970499873161316, + "learning_rate": 9.3209083929076e-06, + "loss": 0.0142, + "step": 7710 + }, + { + "epoch": 12.931323283082078, + "grad_norm": 0.08262616395950317, + "learning_rate": 9.282670454729471e-06, + "loss": 0.0115, + "step": 7720 + }, + { + "epoch": 12.948073701842546, + "grad_norm": 0.055856745690107346, + "learning_rate": 9.244475935868927e-06, + "loss": 0.0117, + "step": 7730 + }, + { + "epoch": 12.964824120603016, + "grad_norm": 0.04862665757536888, + "learning_rate": 9.206325126387945e-06, + "loss": 0.0104, + "step": 7740 + }, + { + "epoch": 12.981574539363484, + "grad_norm": 0.06921498477458954, + "learning_rate": 9.168218316016572e-06, + "loss": 0.0123, + "step": 7750 + }, + { + "epoch": 12.998324958123954, + "grad_norm": 0.09353175759315491, + "learning_rate": 9.130155794150698e-06, + "loss": 0.0121, + "step": 7760 + }, + { + "epoch": 13.015075376884422, + "grad_norm": 0.07180416584014893, + "learning_rate": 9.092137849849876e-06, + "loss": 0.0133, + "step": 7770 + }, + { + "epoch": 13.031825795644892, + "grad_norm": 0.07870851457118988, + "learning_rate": 9.054164771835128e-06, + "loss": 0.0123, + "step": 7780 + }, + { + "epoch": 13.04857621440536, + "grad_norm": 0.06702952831983566, + "learning_rate": 9.016236848486736e-06, + "loss": 0.0107, + "step": 7790 + }, + { + "epoch": 13.06532663316583, + "grad_norm": 0.07762094587087631, + "learning_rate": 8.978354367842068e-06, + "loss": 0.011, + "step": 7800 + }, + { + "epoch": 13.082077051926298, + "grad_norm": 0.07945635914802551, + "learning_rate": 8.940517617593384e-06, + "loss": 0.0119, + "step": 7810 + }, + { + "epoch": 13.098827470686768, + "grad_norm": 0.06439702212810516, + "learning_rate": 8.902726885085655e-06, + "loss": 0.0095, + "step": 7820 + }, + { + "epoch": 13.115577889447236, + "grad_norm": 0.0856538638472557, + "learning_rate": 8.864982457314365e-06, + "loss": 0.0142, + "step": 7830 + }, + { + "epoch": 13.132328308207706, + "grad_norm": 0.07062462717294693, + "learning_rate": 8.827284620923369e-06, + "loss": 0.0126, + "step": 7840 + }, + { + "epoch": 13.149078726968174, + "grad_norm": 0.08214748650789261, + "learning_rate": 8.789633662202667e-06, + "loss": 0.0122, + "step": 7850 + }, + { + "epoch": 13.165829145728644, + "grad_norm": 0.06878134608268738, + "learning_rate": 8.752029867086264e-06, + "loss": 0.0119, + "step": 7860 + }, + { + "epoch": 13.182579564489112, + "grad_norm": 0.10038811713457108, + "learning_rate": 8.714473521149994e-06, + "loss": 0.012, + "step": 7870 + }, + { + "epoch": 13.199329983249582, + "grad_norm": 0.05590192601084709, + "learning_rate": 8.676964909609334e-06, + "loss": 0.0121, + "step": 7880 + }, + { + "epoch": 13.21608040201005, + "grad_norm": 0.08674315363168716, + "learning_rate": 8.639504317317258e-06, + "loss": 0.013, + "step": 7890 + }, + { + "epoch": 13.23283082077052, + "grad_norm": 0.07082588225603104, + "learning_rate": 8.602092028762073e-06, + "loss": 0.0133, + "step": 7900 + }, + { + "epoch": 13.249581239530988, + "grad_norm": 0.09573926776647568, + "learning_rate": 8.56472832806523e-06, + "loss": 0.0137, + "step": 7910 + }, + { + "epoch": 13.266331658291458, + "grad_norm": 0.073761947453022, + "learning_rate": 8.527413498979204e-06, + "loss": 0.0112, + "step": 7920 + }, + { + "epoch": 13.283082077051926, + "grad_norm": 0.06695816665887833, + "learning_rate": 8.490147824885328e-06, + "loss": 0.0112, + "step": 7930 + }, + { + "epoch": 13.299832495812396, + "grad_norm": 0.09559398889541626, + "learning_rate": 8.452931588791614e-06, + "loss": 0.0126, + "step": 7940 + }, + { + "epoch": 13.316582914572864, + "grad_norm": 0.0765177384018898, + "learning_rate": 8.415765073330642e-06, + "loss": 0.0121, + "step": 7950 + }, + { + "epoch": 13.333333333333334, + "grad_norm": 0.06615452468395233, + "learning_rate": 8.378648560757396e-06, + "loss": 0.0131, + "step": 7960 + }, + { + "epoch": 13.350083752093802, + "grad_norm": 0.08720266819000244, + "learning_rate": 8.34158233294711e-06, + "loss": 0.0135, + "step": 7970 + }, + { + "epoch": 13.366834170854272, + "grad_norm": 0.07875142246484756, + "learning_rate": 8.304566671393151e-06, + "loss": 0.0126, + "step": 7980 + }, + { + "epoch": 13.38358458961474, + "grad_norm": 0.04822768643498421, + "learning_rate": 8.267601857204869e-06, + "loss": 0.0119, + "step": 7990 + }, + { + "epoch": 13.40033500837521, + "grad_norm": 0.062463391572237015, + "learning_rate": 8.230688171105446e-06, + "loss": 0.011, + "step": 8000 + }, + { + "epoch": 13.417085427135678, + "grad_norm": 0.06186162680387497, + "learning_rate": 8.193825893429796e-06, + "loss": 0.0108, + "step": 8010 + }, + { + "epoch": 13.433835845896148, + "grad_norm": 0.06208765134215355, + "learning_rate": 8.157015304122424e-06, + "loss": 0.011, + "step": 8020 + }, + { + "epoch": 13.450586264656616, + "grad_norm": 0.06548567116260529, + "learning_rate": 8.120256682735282e-06, + "loss": 0.0111, + "step": 8030 + }, + { + "epoch": 13.467336683417086, + "grad_norm": 0.09333834797143936, + "learning_rate": 8.083550308425658e-06, + "loss": 0.0118, + "step": 8040 + }, + { + "epoch": 13.484087102177554, + "grad_norm": 0.08424598723649979, + "learning_rate": 8.04689645995407e-06, + "loss": 0.0115, + "step": 8050 + }, + { + "epoch": 13.500837520938024, + "grad_norm": 0.08424214273691177, + "learning_rate": 8.010295415682144e-06, + "loss": 0.0113, + "step": 8060 + }, + { + "epoch": 13.517587939698492, + "grad_norm": 0.05681629851460457, + "learning_rate": 7.97374745357047e-06, + "loss": 0.012, + "step": 8070 + }, + { + "epoch": 13.534338358458962, + "grad_norm": 0.09238583594560623, + "learning_rate": 7.937252851176529e-06, + "loss": 0.0124, + "step": 8080 + }, + { + "epoch": 13.55108877721943, + "grad_norm": 0.06826833635568619, + "learning_rate": 7.900811885652577e-06, + "loss": 0.0124, + "step": 8090 + }, + { + "epoch": 13.5678391959799, + "grad_norm": 0.06261388212442398, + "learning_rate": 7.864424833743515e-06, + "loss": 0.0114, + "step": 8100 + }, + { + "epoch": 13.584589614740368, + "grad_norm": 0.056088827550411224, + "learning_rate": 7.828091971784816e-06, + "loss": 0.0111, + "step": 8110 + }, + { + "epoch": 13.601340033500838, + "grad_norm": 0.09993434697389603, + "learning_rate": 7.791813575700425e-06, + "loss": 0.012, + "step": 8120 + }, + { + "epoch": 13.618090452261306, + "grad_norm": 0.04731757566332817, + "learning_rate": 7.755589921000634e-06, + "loss": 0.0128, + "step": 8130 + }, + { + "epoch": 13.634840871021776, + "grad_norm": 0.0865129604935646, + "learning_rate": 7.71942128278003e-06, + "loss": 0.0112, + "step": 8140 + }, + { + "epoch": 13.651591289782244, + "grad_norm": 0.07481952011585236, + "learning_rate": 7.683307935715383e-06, + "loss": 0.0124, + "step": 8150 + }, + { + "epoch": 13.668341708542714, + "grad_norm": 0.07770776748657227, + "learning_rate": 7.647250154063548e-06, + "loss": 0.0132, + "step": 8160 + }, + { + "epoch": 13.685092127303182, + "grad_norm": 0.08305969834327698, + "learning_rate": 7.611248211659417e-06, + "loss": 0.0129, + "step": 8170 + }, + { + "epoch": 13.701842546063652, + "grad_norm": 0.0771985650062561, + "learning_rate": 7.575302381913813e-06, + "loss": 0.0101, + "step": 8180 + }, + { + "epoch": 13.71859296482412, + "grad_norm": 0.06562207639217377, + "learning_rate": 7.539412937811414e-06, + "loss": 0.0116, + "step": 8190 + }, + { + "epoch": 13.73534338358459, + "grad_norm": 0.05711517855525017, + "learning_rate": 7.503580151908694e-06, + "loss": 0.0126, + "step": 8200 + }, + { + "epoch": 13.752093802345058, + "grad_norm": 0.05801519751548767, + "learning_rate": 7.467804296331849e-06, + "loss": 0.0109, + "step": 8210 + }, + { + "epoch": 13.768844221105528, + "grad_norm": 0.08770089596509933, + "learning_rate": 7.432085642774713e-06, + "loss": 0.0109, + "step": 8220 + }, + { + "epoch": 13.785594639865996, + "grad_norm": 0.05994917079806328, + "learning_rate": 7.396424462496722e-06, + "loss": 0.0099, + "step": 8230 + }, + { + "epoch": 13.802345058626466, + "grad_norm": 0.08041512221097946, + "learning_rate": 7.360821026320844e-06, + "loss": 0.0121, + "step": 8240 + }, + { + "epoch": 13.819095477386934, + "grad_norm": 0.0712173655629158, + "learning_rate": 7.325275604631496e-06, + "loss": 0.0127, + "step": 8250 + }, + { + "epoch": 13.835845896147404, + "grad_norm": 0.07469300925731659, + "learning_rate": 7.289788467372536e-06, + "loss": 0.0111, + "step": 8260 + }, + { + "epoch": 13.852596314907872, + "grad_norm": 0.07342658936977386, + "learning_rate": 7.25435988404519e-06, + "loss": 0.0109, + "step": 8270 + }, + { + "epoch": 13.869346733668342, + "grad_norm": 0.08993782103061676, + "learning_rate": 7.218990123705987e-06, + "loss": 0.0123, + "step": 8280 + }, + { + "epoch": 13.88609715242881, + "grad_norm": 0.06399506330490112, + "learning_rate": 7.183679454964752e-06, + "loss": 0.0111, + "step": 8290 + }, + { + "epoch": 13.90284757118928, + "grad_norm": 0.08531822264194489, + "learning_rate": 7.148428145982548e-06, + "loss": 0.0113, + "step": 8300 + }, + { + "epoch": 13.919597989949748, + "grad_norm": 0.06432479619979858, + "learning_rate": 7.113236464469625e-06, + "loss": 0.0113, + "step": 8310 + }, + { + "epoch": 13.936348408710218, + "grad_norm": 0.06948276609182358, + "learning_rate": 7.078104677683417e-06, + "loss": 0.0127, + "step": 8320 + }, + { + "epoch": 13.953098827470686, + "grad_norm": 0.07048949599266052, + "learning_rate": 7.043033052426498e-06, + "loss": 0.0124, + "step": 8330 + }, + { + "epoch": 13.969849246231156, + "grad_norm": 0.11620538681745529, + "learning_rate": 7.00802185504454e-06, + "loss": 0.0119, + "step": 8340 + }, + { + "epoch": 13.986599664991624, + "grad_norm": 0.06149009242653847, + "learning_rate": 6.973071351424319e-06, + "loss": 0.0107, + "step": 8350 + }, + { + "epoch": 14.003350083752094, + "grad_norm": 0.061725832521915436, + "learning_rate": 6.9381818069916835e-06, + "loss": 0.0105, + "step": 8360 + }, + { + "epoch": 14.020100502512562, + "grad_norm": 0.09621080011129379, + "learning_rate": 6.903353486709525e-06, + "loss": 0.0123, + "step": 8370 + }, + { + "epoch": 14.036850921273032, + "grad_norm": 0.05495775490999222, + "learning_rate": 6.868586655075787e-06, + "loss": 0.0091, + "step": 8380 + }, + { + "epoch": 14.0536013400335, + "grad_norm": 0.06942631304264069, + "learning_rate": 6.833881576121452e-06, + "loss": 0.0133, + "step": 8390 + }, + { + "epoch": 14.07035175879397, + "grad_norm": 0.07615401595830917, + "learning_rate": 6.7992385134085175e-06, + "loss": 0.0108, + "step": 8400 + }, + { + "epoch": 14.087102177554438, + "grad_norm": 0.08464915305376053, + "learning_rate": 6.764657730028022e-06, + "loss": 0.0137, + "step": 8410 + }, + { + "epoch": 14.103852596314908, + "grad_norm": 0.07944124191999435, + "learning_rate": 6.7301394885980335e-06, + "loss": 0.0112, + "step": 8420 + }, + { + "epoch": 14.120603015075377, + "grad_norm": 0.07688767462968826, + "learning_rate": 6.69568405126165e-06, + "loss": 0.0106, + "step": 8430 + }, + { + "epoch": 14.137353433835846, + "grad_norm": 0.06504269689321518, + "learning_rate": 6.661291679685004e-06, + "loss": 0.0105, + "step": 8440 + }, + { + "epoch": 14.154103852596315, + "grad_norm": 0.06587643921375275, + "learning_rate": 6.626962635055308e-06, + "loss": 0.0134, + "step": 8450 + }, + { + "epoch": 14.170854271356784, + "grad_norm": 0.07884813100099564, + "learning_rate": 6.5926971780788394e-06, + "loss": 0.0132, + "step": 8460 + }, + { + "epoch": 14.187604690117253, + "grad_norm": 0.04981286823749542, + "learning_rate": 6.558495568978959e-06, + "loss": 0.0109, + "step": 8470 + }, + { + "epoch": 14.204355108877722, + "grad_norm": 0.08177836984395981, + "learning_rate": 6.5243580674941575e-06, + "loss": 0.0104, + "step": 8480 + }, + { + "epoch": 14.22110552763819, + "grad_norm": 0.07541631907224655, + "learning_rate": 6.490284932876075e-06, + "loss": 0.0123, + "step": 8490 + }, + { + "epoch": 14.23785594639866, + "grad_norm": 0.06907821446657181, + "learning_rate": 6.4562764238875085e-06, + "loss": 0.0113, + "step": 8500 + }, + { + "epoch": 14.254606365159129, + "grad_norm": 0.059419356286525726, + "learning_rate": 6.4223327988004835e-06, + "loss": 0.0116, + "step": 8510 + }, + { + "epoch": 14.271356783919598, + "grad_norm": 0.062080204486846924, + "learning_rate": 6.388454315394275e-06, + "loss": 0.0102, + "step": 8520 + }, + { + "epoch": 14.288107202680067, + "grad_norm": 0.07892616838216782, + "learning_rate": 6.3546412309534355e-06, + "loss": 0.0106, + "step": 8530 + }, + { + "epoch": 14.304857621440537, + "grad_norm": 0.08079873770475388, + "learning_rate": 6.32089380226587e-06, + "loss": 0.0131, + "step": 8540 + }, + { + "epoch": 14.321608040201005, + "grad_norm": 0.06901709735393524, + "learning_rate": 6.287212285620876e-06, + "loss": 0.0113, + "step": 8550 + }, + { + "epoch": 14.338358458961475, + "grad_norm": 0.07588738203048706, + "learning_rate": 6.25359693680717e-06, + "loss": 0.012, + "step": 8560 + }, + { + "epoch": 14.355108877721943, + "grad_norm": 0.08524017035961151, + "learning_rate": 6.22004801111099e-06, + "loss": 0.015, + "step": 8570 + }, + { + "epoch": 14.371859296482413, + "grad_norm": 0.0674111470580101, + "learning_rate": 6.186565763314131e-06, + "loss": 0.0111, + "step": 8580 + }, + { + "epoch": 14.38860971524288, + "grad_norm": 0.0773531049489975, + "learning_rate": 6.153150447691996e-06, + "loss": 0.0107, + "step": 8590 + }, + { + "epoch": 14.40536013400335, + "grad_norm": 0.05608673393726349, + "learning_rate": 6.119802318011707e-06, + "loss": 0.0115, + "step": 8600 + }, + { + "epoch": 14.422110552763819, + "grad_norm": 0.06338393688201904, + "learning_rate": 6.0865216275301425e-06, + "loss": 0.0122, + "step": 8610 + }, + { + "epoch": 14.438860971524289, + "grad_norm": 0.074888676404953, + "learning_rate": 6.0533086289920176e-06, + "loss": 0.0106, + "step": 8620 + }, + { + "epoch": 14.455611390284757, + "grad_norm": 0.08034703880548477, + "learning_rate": 6.020163574627984e-06, + "loss": 0.0133, + "step": 8630 + }, + { + "epoch": 14.472361809045227, + "grad_norm": 0.0592302568256855, + "learning_rate": 5.987086716152702e-06, + "loss": 0.011, + "step": 8640 + }, + { + "epoch": 14.489112227805695, + "grad_norm": 0.0759633481502533, + "learning_rate": 5.954078304762916e-06, + "loss": 0.012, + "step": 8650 + }, + { + "epoch": 14.505862646566165, + "grad_norm": 0.0660175308585167, + "learning_rate": 5.921138591135575e-06, + "loss": 0.012, + "step": 8660 + }, + { + "epoch": 14.522613065326633, + "grad_norm": 0.08975532650947571, + "learning_rate": 5.888267825425917e-06, + "loss": 0.0121, + "step": 8670 + }, + { + "epoch": 14.539363484087103, + "grad_norm": 0.07022812962532043, + "learning_rate": 5.8554662572655414e-06, + "loss": 0.0131, + "step": 8680 + }, + { + "epoch": 14.55611390284757, + "grad_norm": 0.057023871690034866, + "learning_rate": 5.822734135760566e-06, + "loss": 0.0119, + "step": 8690 + }, + { + "epoch": 14.57286432160804, + "grad_norm": 0.05912158265709877, + "learning_rate": 5.790071709489698e-06, + "loss": 0.0096, + "step": 8700 + }, + { + "epoch": 14.589614740368509, + "grad_norm": 0.07548556476831436, + "learning_rate": 5.757479226502347e-06, + "loss": 0.0117, + "step": 8710 + }, + { + "epoch": 14.606365159128979, + "grad_norm": 0.07981330156326294, + "learning_rate": 5.724956934316762e-06, + "loss": 0.0132, + "step": 8720 + }, + { + "epoch": 14.623115577889447, + "grad_norm": 0.07118304818868637, + "learning_rate": 5.692505079918138e-06, + "loss": 0.0113, + "step": 8730 + }, + { + "epoch": 14.639865996649917, + "grad_norm": 0.07005375623703003, + "learning_rate": 5.660123909756726e-06, + "loss": 0.011, + "step": 8740 + }, + { + "epoch": 14.656616415410385, + "grad_norm": 0.04754802584648132, + "learning_rate": 5.627813669746001e-06, + "loss": 0.012, + "step": 8750 + }, + { + "epoch": 14.673366834170855, + "grad_norm": 0.09637465327978134, + "learning_rate": 5.595574605260758e-06, + "loss": 0.0103, + "step": 8760 + }, + { + "epoch": 14.690117252931323, + "grad_norm": 0.09745600819587708, + "learning_rate": 5.563406961135256e-06, + "loss": 0.0129, + "step": 8770 + }, + { + "epoch": 14.706867671691793, + "grad_norm": 0.07260012626647949, + "learning_rate": 5.531310981661374e-06, + "loss": 0.0121, + "step": 8780 + }, + { + "epoch": 14.72361809045226, + "grad_norm": 0.07536528259515762, + "learning_rate": 5.499286910586749e-06, + "loss": 0.0118, + "step": 8790 + }, + { + "epoch": 14.74036850921273, + "grad_norm": 0.10453300923109055, + "learning_rate": 5.467334991112902e-06, + "loss": 0.012, + "step": 8800 + }, + { + "epoch": 14.757118927973199, + "grad_norm": 0.06679607182741165, + "learning_rate": 5.435455465893427e-06, + "loss": 0.01, + "step": 8810 + }, + { + "epoch": 14.773869346733669, + "grad_norm": 0.07301633059978485, + "learning_rate": 5.403648577032135e-06, + "loss": 0.0119, + "step": 8820 + }, + { + "epoch": 14.790619765494137, + "grad_norm": 0.08366880565881729, + "learning_rate": 5.3719145660812e-06, + "loss": 0.0115, + "step": 8830 + }, + { + "epoch": 14.807370184254607, + "grad_norm": 0.06086117401719093, + "learning_rate": 5.340253674039336e-06, + "loss": 0.0131, + "step": 8840 + }, + { + "epoch": 14.824120603015075, + "grad_norm": 0.08130698651075363, + "learning_rate": 5.308666141349976e-06, + "loss": 0.0114, + "step": 8850 + }, + { + "epoch": 14.840871021775545, + "grad_norm": 0.06073309853672981, + "learning_rate": 5.277152207899443e-06, + "loss": 0.0113, + "step": 8860 + }, + { + "epoch": 14.857621440536013, + "grad_norm": 0.08015117794275284, + "learning_rate": 5.245712113015101e-06, + "loss": 0.0126, + "step": 8870 + }, + { + "epoch": 14.874371859296483, + "grad_norm": 0.0700366273522377, + "learning_rate": 5.2143460954635805e-06, + "loss": 0.0113, + "step": 8880 + }, + { + "epoch": 14.891122278056951, + "grad_norm": 0.07298661768436432, + "learning_rate": 5.183054393448937e-06, + "loss": 0.0126, + "step": 8890 + }, + { + "epoch": 14.90787269681742, + "grad_norm": 0.055050741881132126, + "learning_rate": 5.151837244610842e-06, + "loss": 0.0107, + "step": 8900 + }, + { + "epoch": 14.924623115577889, + "grad_norm": 0.09468694031238556, + "learning_rate": 5.120694886022792e-06, + "loss": 0.0127, + "step": 8910 + }, + { + "epoch": 14.941373534338359, + "grad_norm": 0.06892071664333344, + "learning_rate": 5.089627554190307e-06, + "loss": 0.0121, + "step": 8920 + }, + { + "epoch": 14.958123953098827, + "grad_norm": 0.0604483038187027, + "learning_rate": 5.058635485049113e-06, + "loss": 0.0142, + "step": 8930 + }, + { + "epoch": 14.974874371859297, + "grad_norm": 0.08353114128112793, + "learning_rate": 5.027718913963378e-06, + "loss": 0.0098, + "step": 8940 + }, + { + "epoch": 14.991624790619765, + "grad_norm": 0.08845911175012589, + "learning_rate": 4.996878075723912e-06, + "loss": 0.0112, + "step": 8950 + }, + { + "epoch": 15.008375209380235, + "grad_norm": 0.059438783675432205, + "learning_rate": 4.966113204546377e-06, + "loss": 0.011, + "step": 8960 + }, + { + "epoch": 15.025125628140703, + "grad_norm": 0.0784023180603981, + "learning_rate": 4.93542453406952e-06, + "loss": 0.013, + "step": 8970 + }, + { + "epoch": 15.041876046901173, + "grad_norm": 0.08503511548042297, + "learning_rate": 4.904812297353403e-06, + "loss": 0.0118, + "step": 8980 + }, + { + "epoch": 15.058626465661641, + "grad_norm": 0.08692113310098648, + "learning_rate": 4.874276726877604e-06, + "loss": 0.0142, + "step": 8990 + }, + { + "epoch": 15.075376884422111, + "grad_norm": 0.08364814519882202, + "learning_rate": 4.843818054539489e-06, + "loss": 0.0112, + "step": 9000 + }, + { + "epoch": 15.09212730318258, + "grad_norm": 0.07530484348535538, + "learning_rate": 4.813436511652433e-06, + "loss": 0.0123, + "step": 9010 + }, + { + "epoch": 15.108877721943049, + "grad_norm": 0.0713859349489212, + "learning_rate": 4.7831323289440525e-06, + "loss": 0.0115, + "step": 9020 + }, + { + "epoch": 15.125628140703517, + "grad_norm": 0.06153390556573868, + "learning_rate": 4.752905736554474e-06, + "loss": 0.0118, + "step": 9030 + }, + { + "epoch": 15.142378559463987, + "grad_norm": 0.06609758734703064, + "learning_rate": 4.72275696403458e-06, + "loss": 0.0105, + "step": 9040 + }, + { + "epoch": 15.159128978224455, + "grad_norm": 0.05786735191941261, + "learning_rate": 4.6926862403442495e-06, + "loss": 0.0103, + "step": 9050 + }, + { + "epoch": 15.175879396984925, + "grad_norm": 0.06739120185375214, + "learning_rate": 4.662693793850643e-06, + "loss": 0.0111, + "step": 9060 + }, + { + "epoch": 15.192629815745393, + "grad_norm": 0.06687992811203003, + "learning_rate": 4.632779852326465e-06, + "loss": 0.011, + "step": 9070 + }, + { + "epoch": 15.209380234505863, + "grad_norm": 0.11040312796831131, + "learning_rate": 4.6029446429482055e-06, + "loss": 0.0124, + "step": 9080 + }, + { + "epoch": 15.226130653266331, + "grad_norm": 0.08872684091329575, + "learning_rate": 4.573188392294452e-06, + "loss": 0.0108, + "step": 9090 + }, + { + "epoch": 15.242881072026801, + "grad_norm": 0.07327628880739212, + "learning_rate": 4.543511326344155e-06, + "loss": 0.013, + "step": 9100 + }, + { + "epoch": 15.25963149078727, + "grad_norm": 0.057703230530023575, + "learning_rate": 4.513913670474896e-06, + "loss": 0.009, + "step": 9110 + }, + { + "epoch": 15.27638190954774, + "grad_norm": 0.0857810378074646, + "learning_rate": 4.4843956494612e-06, + "loss": 0.0115, + "step": 9120 + }, + { + "epoch": 15.293132328308207, + "grad_norm": 0.05844645947217941, + "learning_rate": 4.454957487472821e-06, + "loss": 0.013, + "step": 9130 + }, + { + "epoch": 15.309882747068677, + "grad_norm": 0.07501334697008133, + "learning_rate": 4.4255994080730236e-06, + "loss": 0.0102, + "step": 9140 + }, + { + "epoch": 15.326633165829145, + "grad_norm": 0.09073469042778015, + "learning_rate": 4.396321634216911e-06, + "loss": 0.0116, + "step": 9150 + }, + { + "epoch": 15.343383584589615, + "grad_norm": 0.06694762408733368, + "learning_rate": 4.367124388249717e-06, + "loss": 0.0112, + "step": 9160 + }, + { + "epoch": 15.360134003350083, + "grad_norm": 0.06576213240623474, + "learning_rate": 4.338007891905113e-06, + "loss": 0.0102, + "step": 9170 + }, + { + "epoch": 15.376884422110553, + "grad_norm": 0.05746903270483017, + "learning_rate": 4.308972366303536e-06, + "loss": 0.0093, + "step": 9180 + }, + { + "epoch": 15.393634840871021, + "grad_norm": 0.06974180042743683, + "learning_rate": 4.280018031950506e-06, + "loss": 0.0124, + "step": 9190 + }, + { + "epoch": 15.410385259631491, + "grad_norm": 0.08683093637228012, + "learning_rate": 4.251145108734941e-06, + "loss": 0.0125, + "step": 9200 + }, + { + "epoch": 15.42713567839196, + "grad_norm": 0.05769151821732521, + "learning_rate": 4.222353815927505e-06, + "loss": 0.0113, + "step": 9210 + }, + { + "epoch": 15.44388609715243, + "grad_norm": 0.06937823444604874, + "learning_rate": 4.193644372178931e-06, + "loss": 0.0114, + "step": 9220 + }, + { + "epoch": 15.460636515912897, + "grad_norm": 0.06178734824061394, + "learning_rate": 4.1650169955183535e-06, + "loss": 0.011, + "step": 9230 + }, + { + "epoch": 15.477386934673367, + "grad_norm": 0.055109817534685135, + "learning_rate": 4.136471903351681e-06, + "loss": 0.0099, + "step": 9240 + }, + { + "epoch": 15.494137353433835, + "grad_norm": 0.08306293189525604, + "learning_rate": 4.108009312459902e-06, + "loss": 0.011, + "step": 9250 + }, + { + "epoch": 15.510887772194305, + "grad_norm": 0.07963228225708008, + "learning_rate": 4.079629438997487e-06, + "loss": 0.0128, + "step": 9260 + }, + { + "epoch": 15.527638190954773, + "grad_norm": 0.08196922391653061, + "learning_rate": 4.051332498490699e-06, + "loss": 0.0118, + "step": 9270 + }, + { + "epoch": 15.544388609715243, + "grad_norm": 0.07916904240846634, + "learning_rate": 4.0231187058359996e-06, + "loss": 0.0126, + "step": 9280 + }, + { + "epoch": 15.561139028475711, + "grad_norm": 0.05994735658168793, + "learning_rate": 3.994988275298393e-06, + "loss": 0.0113, + "step": 9290 + }, + { + "epoch": 15.577889447236181, + "grad_norm": 0.07669159024953842, + "learning_rate": 3.966941420509793e-06, + "loss": 0.0118, + "step": 9300 + }, + { + "epoch": 15.59463986599665, + "grad_norm": 0.08148621767759323, + "learning_rate": 3.9389783544674215e-06, + "loss": 0.0113, + "step": 9310 + }, + { + "epoch": 15.61139028475712, + "grad_norm": 0.06318013370037079, + "learning_rate": 3.911099289532183e-06, + "loss": 0.0098, + "step": 9320 + }, + { + "epoch": 15.628140703517587, + "grad_norm": 0.08033207803964615, + "learning_rate": 3.883304437427038e-06, + "loss": 0.0103, + "step": 9330 + }, + { + "epoch": 15.644891122278057, + "grad_norm": 0.06583742052316666, + "learning_rate": 3.855594009235415e-06, + "loss": 0.0133, + "step": 9340 + }, + { + "epoch": 15.661641541038525, + "grad_norm": 0.0734059065580368, + "learning_rate": 3.827968215399601e-06, + "loss": 0.011, + "step": 9350 + }, + { + "epoch": 15.678391959798995, + "grad_norm": 0.059179969131946564, + "learning_rate": 3.800427265719131e-06, + "loss": 0.0124, + "step": 9360 + }, + { + "epoch": 15.695142378559463, + "grad_norm": 0.07946020364761353, + "learning_rate": 3.7729713693492142e-06, + "loss": 0.0103, + "step": 9370 + }, + { + "epoch": 15.711892797319933, + "grad_norm": 0.07202473282814026, + "learning_rate": 3.7456007347991363e-06, + "loss": 0.0107, + "step": 9380 + }, + { + "epoch": 15.728643216080402, + "grad_norm": 0.06295417994260788, + "learning_rate": 3.7183155699306686e-06, + "loss": 0.0104, + "step": 9390 + }, + { + "epoch": 15.745393634840871, + "grad_norm": 0.10039746016263962, + "learning_rate": 3.691116081956502e-06, + "loss": 0.0124, + "step": 9400 + }, + { + "epoch": 15.76214405360134, + "grad_norm": 0.06262555718421936, + "learning_rate": 3.6640024774386754e-06, + "loss": 0.0111, + "step": 9410 + }, + { + "epoch": 15.77889447236181, + "grad_norm": 0.06782974302768707, + "learning_rate": 3.636974962286983e-06, + "loss": 0.0096, + "step": 9420 + }, + { + "epoch": 15.795644891122278, + "grad_norm": 0.07636243849992752, + "learning_rate": 3.610033741757442e-06, + "loss": 0.0106, + "step": 9430 + }, + { + "epoch": 15.812395309882747, + "grad_norm": 0.06166220083832741, + "learning_rate": 3.5831790204507185e-06, + "loss": 0.009, + "step": 9440 + }, + { + "epoch": 15.829145728643216, + "grad_norm": 0.0528729185461998, + "learning_rate": 3.5564110023105602e-06, + "loss": 0.0106, + "step": 9450 + }, + { + "epoch": 15.845896147403685, + "grad_norm": 0.07870613783597946, + "learning_rate": 3.5297298906222796e-06, + "loss": 0.0117, + "step": 9460 + }, + { + "epoch": 15.862646566164154, + "grad_norm": 0.07957785576581955, + "learning_rate": 3.5031358880111873e-06, + "loss": 0.0124, + "step": 9470 + }, + { + "epoch": 15.879396984924623, + "grad_norm": 0.07852274179458618, + "learning_rate": 3.4766291964410474e-06, + "loss": 0.0097, + "step": 9480 + }, + { + "epoch": 15.896147403685092, + "grad_norm": 0.06998380273580551, + "learning_rate": 3.450210017212569e-06, + "loss": 0.012, + "step": 9490 + }, + { + "epoch": 15.912897822445562, + "grad_norm": 0.0631658136844635, + "learning_rate": 3.4238785509618596e-06, + "loss": 0.0087, + "step": 9500 + }, + { + "epoch": 15.92964824120603, + "grad_norm": 0.09431459754705429, + "learning_rate": 3.3976349976588956e-06, + "loss": 0.0107, + "step": 9510 + }, + { + "epoch": 15.9463986599665, + "grad_norm": 0.06583735346794128, + "learning_rate": 3.3714795566060256e-06, + "loss": 0.0119, + "step": 9520 + }, + { + "epoch": 15.963149078726968, + "grad_norm": 0.09137707203626633, + "learning_rate": 3.3454124264364424e-06, + "loss": 0.0095, + "step": 9530 + }, + { + "epoch": 15.979899497487438, + "grad_norm": 0.07819409668445587, + "learning_rate": 3.3194338051126626e-06, + "loss": 0.011, + "step": 9540 + }, + { + "epoch": 15.996649916247906, + "grad_norm": 0.059675078839063644, + "learning_rate": 3.293543889925053e-06, + "loss": 0.0103, + "step": 9550 + }, + { + "epoch": 16.013400335008374, + "grad_norm": 0.08947017788887024, + "learning_rate": 3.2677428774903144e-06, + "loss": 0.0099, + "step": 9560 + }, + { + "epoch": 16.030150753768844, + "grad_norm": 0.06796885281801224, + "learning_rate": 3.2420309637499768e-06, + "loss": 0.0121, + "step": 9570 + }, + { + "epoch": 16.046901172529314, + "grad_norm": 0.10741371661424637, + "learning_rate": 3.2164083439689406e-06, + "loss": 0.0111, + "step": 9580 + }, + { + "epoch": 16.063651591289783, + "grad_norm": 0.08767179399728775, + "learning_rate": 3.1908752127339745e-06, + "loss": 0.0114, + "step": 9590 + }, + { + "epoch": 16.08040201005025, + "grad_norm": 0.0646979808807373, + "learning_rate": 3.1654317639522306e-06, + "loss": 0.0106, + "step": 9600 + }, + { + "epoch": 16.09715242881072, + "grad_norm": 0.07663914561271667, + "learning_rate": 3.140078190849794e-06, + "loss": 0.0107, + "step": 9610 + }, + { + "epoch": 16.11390284757119, + "grad_norm": 0.05584222078323364, + "learning_rate": 3.114814685970204e-06, + "loss": 0.0105, + "step": 9620 + }, + { + "epoch": 16.13065326633166, + "grad_norm": 0.07266643643379211, + "learning_rate": 3.0896414411729784e-06, + "loss": 0.012, + "step": 9630 + }, + { + "epoch": 16.147403685092126, + "grad_norm": 0.12060057371854782, + "learning_rate": 3.064558647632188e-06, + "loss": 0.0114, + "step": 9640 + }, + { + "epoch": 16.164154103852596, + "grad_norm": 0.07776009291410446, + "learning_rate": 3.0395664958349684e-06, + "loss": 0.0116, + "step": 9650 + }, + { + "epoch": 16.180904522613066, + "grad_norm": 0.06058536469936371, + "learning_rate": 3.0146651755801097e-06, + "loss": 0.009, + "step": 9660 + }, + { + "epoch": 16.197654941373536, + "grad_norm": 0.08329997956752777, + "learning_rate": 2.989854875976577e-06, + "loss": 0.01, + "step": 9670 + }, + { + "epoch": 16.214405360134002, + "grad_norm": 0.05222083628177643, + "learning_rate": 2.96513578544211e-06, + "loss": 0.0112, + "step": 9680 + }, + { + "epoch": 16.231155778894472, + "grad_norm": 0.07646694034337997, + "learning_rate": 2.9405080917017747e-06, + "loss": 0.0107, + "step": 9690 + }, + { + "epoch": 16.24790619765494, + "grad_norm": 0.08242721110582352, + "learning_rate": 2.915971981786526e-06, + "loss": 0.0113, + "step": 9700 + }, + { + "epoch": 16.26465661641541, + "grad_norm": 0.06174449622631073, + "learning_rate": 2.891527642031815e-06, + "loss": 0.0115, + "step": 9710 + }, + { + "epoch": 16.281407035175878, + "grad_norm": 0.07003425806760788, + "learning_rate": 2.8671752580761555e-06, + "loss": 0.0098, + "step": 9720 + }, + { + "epoch": 16.298157453936348, + "grad_norm": 0.045007798820734024, + "learning_rate": 2.8429150148597117e-06, + "loss": 0.0101, + "step": 9730 + }, + { + "epoch": 16.314907872696818, + "grad_norm": 0.06540249288082123, + "learning_rate": 2.8187470966229054e-06, + "loss": 0.0111, + "step": 9740 + }, + { + "epoch": 16.331658291457288, + "grad_norm": 0.05390077829360962, + "learning_rate": 2.7946716869050197e-06, + "loss": 0.0141, + "step": 9750 + }, + { + "epoch": 16.348408710217754, + "grad_norm": 0.09093419462442398, + "learning_rate": 2.770688968542777e-06, + "loss": 0.0106, + "step": 9760 + }, + { + "epoch": 16.365159128978224, + "grad_norm": 0.07020417600870132, + "learning_rate": 2.746799123668985e-06, + "loss": 0.0095, + "step": 9770 + }, + { + "epoch": 16.381909547738694, + "grad_norm": 0.0639248639345169, + "learning_rate": 2.7230023337111425e-06, + "loss": 0.0106, + "step": 9780 + }, + { + "epoch": 16.398659966499164, + "grad_norm": 0.07816614955663681, + "learning_rate": 2.6992987793900448e-06, + "loss": 0.0121, + "step": 9790 + }, + { + "epoch": 16.41541038525963, + "grad_norm": 0.059835903346538544, + "learning_rate": 2.6756886407184313e-06, + "loss": 0.0095, + "step": 9800 + }, + { + "epoch": 16.4321608040201, + "grad_norm": 0.07151816040277481, + "learning_rate": 2.6521720969996172e-06, + "loss": 0.0101, + "step": 9810 + }, + { + "epoch": 16.44891122278057, + "grad_norm": 0.09019379317760468, + "learning_rate": 2.6287493268261137e-06, + "loss": 0.0113, + "step": 9820 + }, + { + "epoch": 16.46566164154104, + "grad_norm": 0.08206336200237274, + "learning_rate": 2.6054205080782946e-06, + "loss": 0.0098, + "step": 9830 + }, + { + "epoch": 16.482412060301506, + "grad_norm": 0.06948984414339066, + "learning_rate": 2.582185817923034e-06, + "loss": 0.0112, + "step": 9840 + }, + { + "epoch": 16.499162479061976, + "grad_norm": 0.07291114330291748, + "learning_rate": 2.5590454328123538e-06, + "loss": 0.0102, + "step": 9850 + }, + { + "epoch": 16.515912897822446, + "grad_norm": 0.0553848072886467, + "learning_rate": 2.5359995284820993e-06, + "loss": 0.0118, + "step": 9860 + }, + { + "epoch": 16.532663316582916, + "grad_norm": 0.08315521478652954, + "learning_rate": 2.513048279950602e-06, + "loss": 0.0111, + "step": 9870 + }, + { + "epoch": 16.549413735343382, + "grad_norm": 0.06910669803619385, + "learning_rate": 2.4901918615173274e-06, + "loss": 0.0104, + "step": 9880 + }, + { + "epoch": 16.566164154103852, + "grad_norm": 0.09397564828395844, + "learning_rate": 2.4674304467615876e-06, + "loss": 0.0116, + "step": 9890 + }, + { + "epoch": 16.582914572864322, + "grad_norm": 0.05459596961736679, + "learning_rate": 2.4447642085412007e-06, + "loss": 0.0115, + "step": 9900 + }, + { + "epoch": 16.599664991624792, + "grad_norm": 0.06284655630588531, + "learning_rate": 2.4221933189911734e-06, + "loss": 0.0102, + "step": 9910 + }, + { + "epoch": 16.616415410385258, + "grad_norm": 0.09468510001897812, + "learning_rate": 2.3997179495224104e-06, + "loss": 0.0123, + "step": 9920 + }, + { + "epoch": 16.633165829145728, + "grad_norm": 0.06382662802934647, + "learning_rate": 2.3773382708204096e-06, + "loss": 0.0146, + "step": 9930 + }, + { + "epoch": 16.649916247906198, + "grad_norm": 0.07933798432350159, + "learning_rate": 2.3550544528439443e-06, + "loss": 0.0107, + "step": 9940 + }, + { + "epoch": 16.666666666666668, + "grad_norm": 0.06377198547124863, + "learning_rate": 2.3328666648238076e-06, + "loss": 0.0104, + "step": 9950 + }, + { + "epoch": 16.683417085427134, + "grad_norm": 0.06982989609241486, + "learning_rate": 2.3107750752615027e-06, + "loss": 0.0093, + "step": 9960 + }, + { + "epoch": 16.700167504187604, + "grad_norm": 0.054972875863313675, + "learning_rate": 2.288779851927961e-06, + "loss": 0.0095, + "step": 9970 + }, + { + "epoch": 16.716917922948074, + "grad_norm": 0.06242859736084938, + "learning_rate": 2.266881161862288e-06, + "loss": 0.0107, + "step": 9980 + }, + { + "epoch": 16.733668341708544, + "grad_norm": 0.07019710540771484, + "learning_rate": 2.2450791713704806e-06, + "loss": 0.0102, + "step": 9990 + }, + { + "epoch": 16.75041876046901, + "grad_norm": 0.047752682119607925, + "learning_rate": 2.2233740460241593e-06, + "loss": 0.0111, + "step": 10000 + }, + { + "epoch": 16.76716917922948, + "grad_norm": 0.090115487575531, + "learning_rate": 2.2017659506593256e-06, + "loss": 0.0099, + "step": 10010 + }, + { + "epoch": 16.78391959798995, + "grad_norm": 0.06850699335336685, + "learning_rate": 2.1802550493751043e-06, + "loss": 0.0105, + "step": 10020 + }, + { + "epoch": 16.80067001675042, + "grad_norm": 0.08048104494810104, + "learning_rate": 2.158841505532488e-06, + "loss": 0.0104, + "step": 10030 + }, + { + "epoch": 16.817420435510886, + "grad_norm": 0.06437097489833832, + "learning_rate": 2.1375254817531104e-06, + "loss": 0.0118, + "step": 10040 + }, + { + "epoch": 16.834170854271356, + "grad_norm": 0.07399899512529373, + "learning_rate": 2.1163071399180106e-06, + "loss": 0.0096, + "step": 10050 + }, + { + "epoch": 16.850921273031826, + "grad_norm": 0.08019571006298065, + "learning_rate": 2.0951866411663746e-06, + "loss": 0.0101, + "step": 10060 + }, + { + "epoch": 16.867671691792296, + "grad_norm": 0.06191406771540642, + "learning_rate": 2.0741641458943633e-06, + "loss": 0.0095, + "step": 10070 + }, + { + "epoch": 16.884422110552762, + "grad_norm": 0.06613759696483612, + "learning_rate": 2.053239813753854e-06, + "loss": 0.0112, + "step": 10080 + }, + { + "epoch": 16.901172529313232, + "grad_norm": 0.05650663748383522, + "learning_rate": 2.0324138036512358e-06, + "loss": 0.011, + "step": 10090 + }, + { + "epoch": 16.917922948073702, + "grad_norm": 0.08308195322751999, + "learning_rate": 2.011686273746216e-06, + "loss": 0.011, + "step": 10100 + }, + { + "epoch": 16.934673366834172, + "grad_norm": 0.08628930151462555, + "learning_rate": 1.991057381450608e-06, + "loss": 0.0128, + "step": 10110 + }, + { + "epoch": 16.95142378559464, + "grad_norm": 0.07573705166578293, + "learning_rate": 1.9705272834271337e-06, + "loss": 0.0114, + "step": 10120 + }, + { + "epoch": 16.96817420435511, + "grad_norm": 0.056926365941762924, + "learning_rate": 1.9500961355882418e-06, + "loss": 0.0097, + "step": 10130 + }, + { + "epoch": 16.984924623115578, + "grad_norm": 0.06239789351820946, + "learning_rate": 1.9297640930949217e-06, + "loss": 0.0098, + "step": 10140 + }, + { + "epoch": 17.001675041876048, + "grad_norm": 0.06450553983449936, + "learning_rate": 1.909531310355518e-06, + "loss": 0.0085, + "step": 10150 + }, + { + "epoch": 17.018425460636514, + "grad_norm": 0.07283646613359451, + "learning_rate": 1.8893979410245694e-06, + "loss": 0.0104, + "step": 10160 + }, + { + "epoch": 17.035175879396984, + "grad_norm": 0.07477754354476929, + "learning_rate": 1.8693641380016341e-06, + "loss": 0.0106, + "step": 10170 + }, + { + "epoch": 17.051926298157454, + "grad_norm": 0.0838644728064537, + "learning_rate": 1.8494300534301268e-06, + "loss": 0.0119, + "step": 10180 + }, + { + "epoch": 17.068676716917924, + "grad_norm": 0.07156196236610413, + "learning_rate": 1.8295958386961691e-06, + "loss": 0.0096, + "step": 10190 + }, + { + "epoch": 17.08542713567839, + "grad_norm": 0.052184104919433594, + "learning_rate": 1.8098616444274435e-06, + "loss": 0.0104, + "step": 10200 + }, + { + "epoch": 17.10217755443886, + "grad_norm": 0.07056842744350433, + "learning_rate": 1.7902276204920326e-06, + "loss": 0.0108, + "step": 10210 + }, + { + "epoch": 17.11892797319933, + "grad_norm": 0.06949982792139053, + "learning_rate": 1.770693915997298e-06, + "loss": 0.0113, + "step": 10220 + }, + { + "epoch": 17.1356783919598, + "grad_norm": 0.06455013900995255, + "learning_rate": 1.7512606792887436e-06, + "loss": 0.0095, + "step": 10230 + }, + { + "epoch": 17.152428810720266, + "grad_norm": 0.06968142837285995, + "learning_rate": 1.7319280579488877e-06, + "loss": 0.0103, + "step": 10240 + }, + { + "epoch": 17.169179229480736, + "grad_norm": 0.07281049340963364, + "learning_rate": 1.712696198796131e-06, + "loss": 0.0098, + "step": 10250 + }, + { + "epoch": 17.185929648241206, + "grad_norm": 0.0891086608171463, + "learning_rate": 1.6935652478836644e-06, + "loss": 0.011, + "step": 10260 + }, + { + "epoch": 17.202680067001676, + "grad_norm": 0.06211129575967789, + "learning_rate": 1.6745353504983473e-06, + "loss": 0.0108, + "step": 10270 + }, + { + "epoch": 17.219430485762143, + "grad_norm": 0.08370804041624069, + "learning_rate": 1.6556066511595947e-06, + "loss": 0.0105, + "step": 10280 + }, + { + "epoch": 17.236180904522612, + "grad_norm": 0.0769832655787468, + "learning_rate": 1.636779293618299e-06, + "loss": 0.0126, + "step": 10290 + }, + { + "epoch": 17.252931323283082, + "grad_norm": 0.06175370141863823, + "learning_rate": 1.618053420855729e-06, + "loss": 0.0094, + "step": 10300 + }, + { + "epoch": 17.269681742043552, + "grad_norm": 0.08861753344535828, + "learning_rate": 1.5994291750824364e-06, + "loss": 0.0117, + "step": 10310 + }, + { + "epoch": 17.28643216080402, + "grad_norm": 0.057883117347955704, + "learning_rate": 1.58090669773719e-06, + "loss": 0.0096, + "step": 10320 + }, + { + "epoch": 17.30318257956449, + "grad_norm": 0.05262555181980133, + "learning_rate": 1.5624861294858971e-06, + "loss": 0.0098, + "step": 10330 + }, + { + "epoch": 17.31993299832496, + "grad_norm": 0.07395872473716736, + "learning_rate": 1.5441676102205244e-06, + "loss": 0.0103, + "step": 10340 + }, + { + "epoch": 17.33668341708543, + "grad_norm": 0.06096988916397095, + "learning_rate": 1.5259512790580493e-06, + "loss": 0.0105, + "step": 10350 + }, + { + "epoch": 17.353433835845895, + "grad_norm": 0.05924631655216217, + "learning_rate": 1.5078372743394043e-06, + "loss": 0.0104, + "step": 10360 + }, + { + "epoch": 17.370184254606365, + "grad_norm": 0.0643719881772995, + "learning_rate": 1.4898257336284078e-06, + "loss": 0.0114, + "step": 10370 + }, + { + "epoch": 17.386934673366834, + "grad_norm": 0.07372860610485077, + "learning_rate": 1.4719167937107425e-06, + "loss": 0.0093, + "step": 10380 + }, + { + "epoch": 17.403685092127304, + "grad_norm": 0.0642753317952156, + "learning_rate": 1.4541105905929047e-06, + "loss": 0.0102, + "step": 10390 + }, + { + "epoch": 17.42043551088777, + "grad_norm": 0.09747152030467987, + "learning_rate": 1.4364072595011679e-06, + "loss": 0.0134, + "step": 10400 + }, + { + "epoch": 17.43718592964824, + "grad_norm": 0.0734635517001152, + "learning_rate": 1.4188069348805655e-06, + "loss": 0.0102, + "step": 10410 + }, + { + "epoch": 17.45393634840871, + "grad_norm": 0.05024496838450432, + "learning_rate": 1.4013097503938682e-06, + "loss": 0.0107, + "step": 10420 + }, + { + "epoch": 17.47068676716918, + "grad_norm": 0.0604495108127594, + "learning_rate": 1.3839158389205536e-06, + "loss": 0.0096, + "step": 10430 + }, + { + "epoch": 17.487437185929647, + "grad_norm": 0.06331069022417068, + "learning_rate": 1.3666253325558232e-06, + "loss": 0.0084, + "step": 10440 + }, + { + "epoch": 17.504187604690117, + "grad_norm": 0.08753936737775803, + "learning_rate": 1.3494383626095797e-06, + "loss": 0.01, + "step": 10450 + }, + { + "epoch": 17.520938023450586, + "grad_norm": 0.0691802129149437, + "learning_rate": 1.3323550596054368e-06, + "loss": 0.0102, + "step": 10460 + }, + { + "epoch": 17.537688442211056, + "grad_norm": 0.06930302828550339, + "learning_rate": 1.3153755532797196e-06, + "loss": 0.0105, + "step": 10470 + }, + { + "epoch": 17.554438860971523, + "grad_norm": 0.11355336755514145, + "learning_rate": 1.2984999725804986e-06, + "loss": 0.01, + "step": 10480 + }, + { + "epoch": 17.571189279731993, + "grad_norm": 0.08153529465198517, + "learning_rate": 1.2817284456665856e-06, + "loss": 0.0121, + "step": 10490 + }, + { + "epoch": 17.587939698492463, + "grad_norm": 0.05092867091298103, + "learning_rate": 1.2650610999065849e-06, + "loss": 0.011, + "step": 10500 + }, + { + "epoch": 17.604690117252932, + "grad_norm": 0.06719738990068436, + "learning_rate": 1.2484980618779097e-06, + "loss": 0.0111, + "step": 10510 + }, + { + "epoch": 17.6214405360134, + "grad_norm": 0.07338828593492508, + "learning_rate": 1.2320394573658188e-06, + "loss": 0.0096, + "step": 10520 + }, + { + "epoch": 17.63819095477387, + "grad_norm": 0.10995873808860779, + "learning_rate": 1.215685411362476e-06, + "loss": 0.011, + "step": 10530 + }, + { + "epoch": 17.65494137353434, + "grad_norm": 0.060237374156713486, + "learning_rate": 1.1994360480659945e-06, + "loss": 0.0107, + "step": 10540 + }, + { + "epoch": 17.67169179229481, + "grad_norm": 0.07575573772192001, + "learning_rate": 1.18329149087948e-06, + "loss": 0.0121, + "step": 10550 + }, + { + "epoch": 17.688442211055275, + "grad_norm": 0.05417560040950775, + "learning_rate": 1.1672518624101176e-06, + "loss": 0.0099, + "step": 10560 + }, + { + "epoch": 17.705192629815745, + "grad_norm": 0.05768902972340584, + "learning_rate": 1.1513172844682251e-06, + "loss": 0.0096, + "step": 10570 + }, + { + "epoch": 17.721943048576215, + "grad_norm": 0.07235273718833923, + "learning_rate": 1.1354878780663263e-06, + "loss": 0.0115, + "step": 10580 + }, + { + "epoch": 17.738693467336685, + "grad_norm": 0.06070183590054512, + "learning_rate": 1.1197637634182433e-06, + "loss": 0.0111, + "step": 10590 + }, + { + "epoch": 17.75544388609715, + "grad_norm": 0.06319398432970047, + "learning_rate": 1.1041450599381765e-06, + "loss": 0.0087, + "step": 10600 + }, + { + "epoch": 17.77219430485762, + "grad_norm": 0.06955882161855698, + "learning_rate": 1.0886318862397926e-06, + "loss": 0.0098, + "step": 10610 + }, + { + "epoch": 17.78894472361809, + "grad_norm": 0.07007668912410736, + "learning_rate": 1.073224360135337e-06, + "loss": 0.0124, + "step": 10620 + }, + { + "epoch": 17.80569514237856, + "grad_norm": 0.06133275106549263, + "learning_rate": 1.0579225986347312e-06, + "loss": 0.0105, + "step": 10630 + }, + { + "epoch": 17.822445561139027, + "grad_norm": 0.05876230075955391, + "learning_rate": 1.042726717944678e-06, + "loss": 0.0128, + "step": 10640 + }, + { + "epoch": 17.839195979899497, + "grad_norm": 0.06510904431343079, + "learning_rate": 1.0276368334677932e-06, + "loss": 0.0105, + "step": 10650 + }, + { + "epoch": 17.855946398659967, + "grad_norm": 0.08215858787298203, + "learning_rate": 1.0126530598017215e-06, + "loss": 0.0099, + "step": 10660 + }, + { + "epoch": 17.872696817420437, + "grad_norm": 0.07347946614027023, + "learning_rate": 9.977755107382586e-07, + "loss": 0.0094, + "step": 10670 + }, + { + "epoch": 17.889447236180903, + "grad_norm": 0.05123291164636612, + "learning_rate": 9.830042992625027e-07, + "loss": 0.0106, + "step": 10680 + }, + { + "epoch": 17.906197654941373, + "grad_norm": 0.06298977136611938, + "learning_rate": 9.683395375519911e-07, + "loss": 0.0116, + "step": 10690 + }, + { + "epoch": 17.922948073701843, + "grad_norm": 0.08945272117853165, + "learning_rate": 9.537813369758359e-07, + "loss": 0.0108, + "step": 10700 + }, + { + "epoch": 17.939698492462313, + "grad_norm": 0.08734112977981567, + "learning_rate": 9.393298080938934e-07, + "loss": 0.0105, + "step": 10710 + }, + { + "epoch": 17.95644891122278, + "grad_norm": 0.08413497358560562, + "learning_rate": 9.249850606559246e-07, + "loss": 0.0105, + "step": 10720 + }, + { + "epoch": 17.97319932998325, + "grad_norm": 0.06129211187362671, + "learning_rate": 9.107472036007458e-07, + "loss": 0.0117, + "step": 10730 + }, + { + "epoch": 17.98994974874372, + "grad_norm": 0.05145135521888733, + "learning_rate": 8.96616345055421e-07, + "loss": 0.01, + "step": 10740 + }, + { + "epoch": 18.00670016750419, + "grad_norm": 0.04860294237732887, + "learning_rate": 8.825925923344275e-07, + "loss": 0.011, + "step": 10750 + }, + { + "epoch": 18.023450586264655, + "grad_norm": 0.06857185065746307, + "learning_rate": 8.686760519388432e-07, + "loss": 0.0085, + "step": 10760 + }, + { + "epoch": 18.040201005025125, + "grad_norm": 0.051961518824100494, + "learning_rate": 8.548668295555439e-07, + "loss": 0.0096, + "step": 10770 + }, + { + "epoch": 18.056951423785595, + "grad_norm": 0.0770072340965271, + "learning_rate": 8.411650300563961e-07, + "loss": 0.0091, + "step": 10780 + }, + { + "epoch": 18.073701842546065, + "grad_norm": 0.06823989003896713, + "learning_rate": 8.275707574974551e-07, + "loss": 0.0091, + "step": 10790 + }, + { + "epoch": 18.09045226130653, + "grad_norm": 0.08679325133562088, + "learning_rate": 8.140841151181894e-07, + "loss": 0.012, + "step": 10800 + }, + { + "epoch": 18.107202680067, + "grad_norm": 0.0784691721200943, + "learning_rate": 8.007052053406816e-07, + "loss": 0.0116, + "step": 10810 + }, + { + "epoch": 18.12395309882747, + "grad_norm": 0.07049158215522766, + "learning_rate": 7.87434129768857e-07, + "loss": 0.0096, + "step": 10820 + }, + { + "epoch": 18.14070351758794, + "grad_norm": 0.06313124299049377, + "learning_rate": 7.742709891877159e-07, + "loss": 0.0103, + "step": 10830 + }, + { + "epoch": 18.157453936348407, + "grad_norm": 0.09241461008787155, + "learning_rate": 7.61215883562561e-07, + "loss": 0.0099, + "step": 10840 + }, + { + "epoch": 18.174204355108877, + "grad_norm": 0.06502427905797958, + "learning_rate": 7.482689120382397e-07, + "loss": 0.0112, + "step": 10850 + }, + { + "epoch": 18.190954773869347, + "grad_norm": 0.076650470495224, + "learning_rate": 7.354301729383944e-07, + "loss": 0.011, + "step": 10860 + }, + { + "epoch": 18.207705192629817, + "grad_norm": 0.07330027967691422, + "learning_rate": 7.226997637647204e-07, + "loss": 0.0107, + "step": 10870 + }, + { + "epoch": 18.224455611390283, + "grad_norm": 0.05841333046555519, + "learning_rate": 7.100777811962072e-07, + "loss": 0.0107, + "step": 10880 + }, + { + "epoch": 18.241206030150753, + "grad_norm": 0.07078751176595688, + "learning_rate": 6.975643210884252e-07, + "loss": 0.0098, + "step": 10890 + }, + { + "epoch": 18.257956448911223, + "grad_norm": 0.05953970178961754, + "learning_rate": 6.851594784727838e-07, + "loss": 0.0106, + "step": 10900 + }, + { + "epoch": 18.274706867671693, + "grad_norm": 0.0576588436961174, + "learning_rate": 6.728633475558221e-07, + "loss": 0.0112, + "step": 10910 + }, + { + "epoch": 18.29145728643216, + "grad_norm": 0.058967769145965576, + "learning_rate": 6.606760217184766e-07, + "loss": 0.0115, + "step": 10920 + }, + { + "epoch": 18.30820770519263, + "grad_norm": 0.057810813188552856, + "learning_rate": 6.485975935153877e-07, + "loss": 0.0109, + "step": 10930 + }, + { + "epoch": 18.3249581239531, + "grad_norm": 0.07053758949041367, + "learning_rate": 6.366281546741909e-07, + "loss": 0.0103, + "step": 10940 + }, + { + "epoch": 18.34170854271357, + "grad_norm": 0.06809135526418686, + "learning_rate": 6.247677960948139e-07, + "loss": 0.0103, + "step": 10950 + }, + { + "epoch": 18.358458961474035, + "grad_norm": 0.05694377422332764, + "learning_rate": 6.130166078487987e-07, + "loss": 0.0111, + "step": 10960 + }, + { + "epoch": 18.375209380234505, + "grad_norm": 0.09857413917779922, + "learning_rate": 6.013746791786117e-07, + "loss": 0.011, + "step": 10970 + }, + { + "epoch": 18.391959798994975, + "grad_norm": 0.07620888203382492, + "learning_rate": 5.898420984969588e-07, + "loss": 0.0103, + "step": 10980 + }, + { + "epoch": 18.408710217755445, + "grad_norm": 0.06404457986354828, + "learning_rate": 5.784189533861262e-07, + "loss": 0.0086, + "step": 10990 + }, + { + "epoch": 18.42546063651591, + "grad_norm": 0.06898780912160873, + "learning_rate": 5.671053305973084e-07, + "loss": 0.0134, + "step": 11000 + }, + { + "epoch": 18.44221105527638, + "grad_norm": 0.0845097228884697, + "learning_rate": 5.559013160499504e-07, + "loss": 0.0107, + "step": 11010 + }, + { + "epoch": 18.45896147403685, + "grad_norm": 0.07644879817962646, + "learning_rate": 5.44806994831093e-07, + "loss": 0.0111, + "step": 11020 + }, + { + "epoch": 18.47571189279732, + "grad_norm": 0.0650167316198349, + "learning_rate": 5.338224511947348e-07, + "loss": 0.0103, + "step": 11030 + }, + { + "epoch": 18.492462311557787, + "grad_norm": 0.07331617176532745, + "learning_rate": 5.229477685611799e-07, + "loss": 0.009, + "step": 11040 + }, + { + "epoch": 18.509212730318257, + "grad_norm": 0.05208346247673035, + "learning_rate": 5.121830295164182e-07, + "loss": 0.0102, + "step": 11050 + }, + { + "epoch": 18.525963149078727, + "grad_norm": 0.05402592569589615, + "learning_rate": 5.015283158114853e-07, + "loss": 0.012, + "step": 11060 + }, + { + "epoch": 18.542713567839197, + "grad_norm": 0.05138329043984413, + "learning_rate": 4.90983708361849e-07, + "loss": 0.0092, + "step": 11070 + }, + { + "epoch": 18.559463986599663, + "grad_norm": 0.07261831313371658, + "learning_rate": 4.805492872467942e-07, + "loss": 0.0111, + "step": 11080 + }, + { + "epoch": 18.576214405360133, + "grad_norm": 0.04949916899204254, + "learning_rate": 4.702251317088152e-07, + "loss": 0.0105, + "step": 11090 + }, + { + "epoch": 18.592964824120603, + "grad_norm": 0.0791100412607193, + "learning_rate": 4.600113201530093e-07, + "loss": 0.0105, + "step": 11100 + }, + { + "epoch": 18.609715242881073, + "grad_norm": 0.07842051982879639, + "learning_rate": 4.499079301464859e-07, + "loss": 0.0098, + "step": 11110 + }, + { + "epoch": 18.62646566164154, + "grad_norm": 0.060268741101026535, + "learning_rate": 4.399150384177786e-07, + "loss": 0.0107, + "step": 11120 + }, + { + "epoch": 18.64321608040201, + "grad_norm": 0.07312291860580444, + "learning_rate": 4.3003272085625376e-07, + "loss": 0.0108, + "step": 11130 + }, + { + "epoch": 18.65996649916248, + "grad_norm": 0.07874402403831482, + "learning_rate": 4.20261052511548e-07, + "loss": 0.0107, + "step": 11140 + }, + { + "epoch": 18.67671691792295, + "grad_norm": 0.0655098631978035, + "learning_rate": 4.106001075929883e-07, + "loss": 0.0096, + "step": 11150 + }, + { + "epoch": 18.693467336683415, + "grad_norm": 0.06539636105298996, + "learning_rate": 4.0104995946902424e-07, + "loss": 0.0092, + "step": 11160 + }, + { + "epoch": 18.710217755443885, + "grad_norm": 0.06471475958824158, + "learning_rate": 3.916106806666869e-07, + "loss": 0.0114, + "step": 11170 + }, + { + "epoch": 18.726968174204355, + "grad_norm": 0.047799427062273026, + "learning_rate": 3.8228234287102423e-07, + "loss": 0.0103, + "step": 11180 + }, + { + "epoch": 18.743718592964825, + "grad_norm": 0.05634230747818947, + "learning_rate": 3.730650169245597e-07, + "loss": 0.0102, + "step": 11190 + }, + { + "epoch": 18.76046901172529, + "grad_norm": 0.059011902660131454, + "learning_rate": 3.639587728267579e-07, + "loss": 0.0119, + "step": 11200 + }, + { + "epoch": 18.77721943048576, + "grad_norm": 0.0783902257680893, + "learning_rate": 3.549636797334915e-07, + "loss": 0.013, + "step": 11210 + }, + { + "epoch": 18.79396984924623, + "grad_norm": 0.061978306621313095, + "learning_rate": 3.460798059565101e-07, + "loss": 0.01, + "step": 11220 + }, + { + "epoch": 18.8107202680067, + "grad_norm": 0.07768058031797409, + "learning_rate": 3.3730721896293227e-07, + "loss": 0.0124, + "step": 11230 + }, + { + "epoch": 18.827470686767168, + "grad_norm": 0.07385557144880295, + "learning_rate": 3.286459853747259e-07, + "loss": 0.0108, + "step": 11240 + }, + { + "epoch": 18.844221105527637, + "grad_norm": 0.08369773626327515, + "learning_rate": 3.200961709681988e-07, + "loss": 0.0124, + "step": 11250 + }, + { + "epoch": 18.860971524288107, + "grad_norm": 0.04872307553887367, + "learning_rate": 3.1165784067351213e-07, + "loss": 0.0095, + "step": 11260 + }, + { + "epoch": 18.877721943048577, + "grad_norm": 0.0770203247666359, + "learning_rate": 3.033310585741761e-07, + "loss": 0.0107, + "step": 11270 + }, + { + "epoch": 18.894472361809044, + "grad_norm": 0.07913050055503845, + "learning_rate": 2.951158879065652e-07, + "loss": 0.009, + "step": 11280 + }, + { + "epoch": 18.911222780569513, + "grad_norm": 0.05107510834932327, + "learning_rate": 2.870123910594369e-07, + "loss": 0.0101, + "step": 11290 + }, + { + "epoch": 18.927973199329983, + "grad_norm": 0.08134544640779495, + "learning_rate": 2.790206295734654e-07, + "loss": 0.0087, + "step": 11300 + }, + { + "epoch": 18.944723618090453, + "grad_norm": 0.06199859082698822, + "learning_rate": 2.711406641407688e-07, + "loss": 0.009, + "step": 11310 + }, + { + "epoch": 18.96147403685092, + "grad_norm": 0.10949202626943588, + "learning_rate": 2.633725546044441e-07, + "loss": 0.0091, + "step": 11320 + }, + { + "epoch": 18.97822445561139, + "grad_norm": 0.053913816809654236, + "learning_rate": 2.5571635995812136e-07, + "loss": 0.0105, + "step": 11330 + }, + { + "epoch": 18.99497487437186, + "grad_norm": 0.06924161314964294, + "learning_rate": 2.4817213834551046e-07, + "loss": 0.0094, + "step": 11340 + }, + { + "epoch": 19.01172529313233, + "grad_norm": 0.08277201652526855, + "learning_rate": 2.4073994705995637e-07, + "loss": 0.0118, + "step": 11350 + }, + { + "epoch": 19.028475711892796, + "grad_norm": 0.08757414668798447, + "learning_rate": 2.3341984254401648e-07, + "loss": 0.0116, + "step": 11360 + }, + { + "epoch": 19.045226130653266, + "grad_norm": 0.07003695517778397, + "learning_rate": 2.2621188038901563e-07, + "loss": 0.0108, + "step": 11370 + }, + { + "epoch": 19.061976549413735, + "grad_norm": 0.07919664680957794, + "learning_rate": 2.191161153346366e-07, + "loss": 0.0103, + "step": 11380 + }, + { + "epoch": 19.078726968174205, + "grad_norm": 0.07476326823234558, + "learning_rate": 2.121326012684971e-07, + "loss": 0.0113, + "step": 11390 + }, + { + "epoch": 19.09547738693467, + "grad_norm": 0.04848461598157883, + "learning_rate": 2.052613912257467e-07, + "loss": 0.0104, + "step": 11400 + }, + { + "epoch": 19.11222780569514, + "grad_norm": 0.07294335961341858, + "learning_rate": 1.9850253738865564e-07, + "loss": 0.0116, + "step": 11410 + }, + { + "epoch": 19.12897822445561, + "grad_norm": 0.07061304897069931, + "learning_rate": 1.9185609108622825e-07, + "loss": 0.01, + "step": 11420 + }, + { + "epoch": 19.14572864321608, + "grad_norm": 0.047857094556093216, + "learning_rate": 1.853221027938068e-07, + "loss": 0.0089, + "step": 11430 + }, + { + "epoch": 19.162479061976548, + "grad_norm": 0.07067311555147171, + "learning_rate": 1.7890062213268665e-07, + "loss": 0.0111, + "step": 11440 + }, + { + "epoch": 19.179229480737018, + "grad_norm": 0.08010688424110413, + "learning_rate": 1.7259169786974672e-07, + "loss": 0.0105, + "step": 11450 + }, + { + "epoch": 19.195979899497488, + "grad_norm": 0.06564082950353622, + "learning_rate": 1.663953779170746e-07, + "loss": 0.0113, + "step": 11460 + }, + { + "epoch": 19.212730318257957, + "grad_norm": 0.055230334401130676, + "learning_rate": 1.6031170933159867e-07, + "loss": 0.0101, + "step": 11470 + }, + { + "epoch": 19.229480737018424, + "grad_norm": 0.06035526841878891, + "learning_rate": 1.5434073831473994e-07, + "loss": 0.009, + "step": 11480 + }, + { + "epoch": 19.246231155778894, + "grad_norm": 0.06482507288455963, + "learning_rate": 1.4848251021205405e-07, + "loss": 0.0086, + "step": 11490 + }, + { + "epoch": 19.262981574539364, + "grad_norm": 0.041663289070129395, + "learning_rate": 1.4273706951288989e-07, + "loss": 0.0103, + "step": 11500 + }, + { + "epoch": 19.279731993299833, + "grad_norm": 0.061265233904123306, + "learning_rate": 1.3710445985004815e-07, + "loss": 0.0112, + "step": 11510 + }, + { + "epoch": 19.2964824120603, + "grad_norm": 0.05612942576408386, + "learning_rate": 1.3158472399945996e-07, + "loss": 0.0103, + "step": 11520 + }, + { + "epoch": 19.31323283082077, + "grad_norm": 0.05675517022609711, + "learning_rate": 1.2617790387984385e-07, + "loss": 0.0128, + "step": 11530 + }, + { + "epoch": 19.32998324958124, + "grad_norm": 0.07928426563739777, + "learning_rate": 1.2088404055241088e-07, + "loss": 0.0111, + "step": 11540 + }, + { + "epoch": 19.34673366834171, + "grad_norm": 0.05637091398239136, + "learning_rate": 1.1570317422053e-07, + "loss": 0.0095, + "step": 11550 + }, + { + "epoch": 19.363484087102176, + "grad_norm": 0.08174281567335129, + "learning_rate": 1.1063534422943833e-07, + "loss": 0.0117, + "step": 11560 + }, + { + "epoch": 19.380234505862646, + "grad_norm": 0.09327259659767151, + "learning_rate": 1.0568058906593791e-07, + "loss": 0.0091, + "step": 11570 + }, + { + "epoch": 19.396984924623116, + "grad_norm": 0.04892966151237488, + "learning_rate": 1.0083894635809943e-07, + "loss": 0.0104, + "step": 11580 + }, + { + "epoch": 19.413735343383586, + "grad_norm": 0.08754295855760574, + "learning_rate": 9.611045287497899e-08, + "loss": 0.0121, + "step": 11590 + }, + { + "epoch": 19.430485762144052, + "grad_norm": 0.061934325844049454, + "learning_rate": 9.149514452634511e-08, + "loss": 0.0102, + "step": 11600 + }, + { + "epoch": 19.44723618090452, + "grad_norm": 0.0838724821805954, + "learning_rate": 8.699305636239386e-08, + "loss": 0.012, + "step": 11610 + }, + { + "epoch": 19.46398659966499, + "grad_norm": 0.07911603897809982, + "learning_rate": 8.260422257349409e-08, + "loss": 0.0133, + "step": 11620 + }, + { + "epoch": 19.48073701842546, + "grad_norm": 0.09958657622337341, + "learning_rate": 7.832867648992104e-08, + "loss": 0.011, + "step": 11630 + }, + { + "epoch": 19.497487437185928, + "grad_norm": 0.08941280841827393, + "learning_rate": 7.416645058160642e-08, + "loss": 0.0115, + "step": 11640 + }, + { + "epoch": 19.514237855946398, + "grad_norm": 0.06926604360342026, + "learning_rate": 7.011757645788541e-08, + "loss": 0.0115, + "step": 11650 + }, + { + "epoch": 19.530988274706868, + "grad_norm": 0.06882479041814804, + "learning_rate": 6.618208486727006e-08, + "loss": 0.0094, + "step": 11660 + }, + { + "epoch": 19.547738693467338, + "grad_norm": 0.052707117050886154, + "learning_rate": 6.236000569719957e-08, + "loss": 0.0101, + "step": 11670 + }, + { + "epoch": 19.564489112227804, + "grad_norm": 0.055740345269441605, + "learning_rate": 5.865136797382875e-08, + "loss": 0.0102, + "step": 11680 + }, + { + "epoch": 19.581239530988274, + "grad_norm": 0.04945455119013786, + "learning_rate": 5.505619986179156e-08, + "loss": 0.0087, + "step": 11690 + }, + { + "epoch": 19.597989949748744, + "grad_norm": 0.07951685786247253, + "learning_rate": 5.157452866400292e-08, + "loss": 0.0121, + "step": 11700 + }, + { + "epoch": 19.614740368509214, + "grad_norm": 0.05180259048938751, + "learning_rate": 4.820638082144058e-08, + "loss": 0.0103, + "step": 11710 + }, + { + "epoch": 19.631490787269684, + "grad_norm": 0.03335301950573921, + "learning_rate": 4.495178191294691e-08, + "loss": 0.0094, + "step": 11720 + }, + { + "epoch": 19.64824120603015, + "grad_norm": 0.032290928065776825, + "learning_rate": 4.181075665504075e-08, + "loss": 0.0092, + "step": 11730 + }, + { + "epoch": 19.66499162479062, + "grad_norm": 0.07559007406234741, + "learning_rate": 3.8783328901719203e-08, + "loss": 0.0113, + "step": 11740 + }, + { + "epoch": 19.68174204355109, + "grad_norm": 0.06876268982887268, + "learning_rate": 3.586952164428281e-08, + "loss": 0.0114, + "step": 11750 + }, + { + "epoch": 19.698492462311556, + "grad_norm": 0.06580738723278046, + "learning_rate": 3.3069357011162314e-08, + "loss": 0.0103, + "step": 11760 + }, + { + "epoch": 19.715242881072026, + "grad_norm": 0.06227555125951767, + "learning_rate": 3.038285626774551e-08, + "loss": 0.0105, + "step": 11770 + }, + { + "epoch": 19.731993299832496, + "grad_norm": 0.07759340107440948, + "learning_rate": 2.781003981622232e-08, + "loss": 0.0084, + "step": 11780 + }, + { + "epoch": 19.748743718592966, + "grad_norm": 0.0702289342880249, + "learning_rate": 2.535092719541998e-08, + "loss": 0.0099, + "step": 11790 + }, + { + "epoch": 19.765494137353436, + "grad_norm": 0.07710360735654831, + "learning_rate": 2.3005537080668105e-08, + "loss": 0.0113, + "step": 11800 + }, + { + "epoch": 19.782244556113902, + "grad_norm": 0.09792066365480423, + "learning_rate": 2.0773887283642156e-08, + "loss": 0.0106, + "step": 11810 + }, + { + "epoch": 19.798994974874372, + "grad_norm": 0.04683520272374153, + "learning_rate": 1.865599475224189e-08, + "loss": 0.0097, + "step": 11820 + }, + { + "epoch": 19.81574539363484, + "grad_norm": 0.06832157075405121, + "learning_rate": 1.6651875570451447e-08, + "loss": 0.0124, + "step": 11830 + }, + { + "epoch": 19.832495812395308, + "grad_norm": 0.06539963185787201, + "learning_rate": 1.4761544958222794e-08, + "loss": 0.0117, + "step": 11840 + }, + { + "epoch": 19.849246231155778, + "grad_norm": 0.06669726967811584, + "learning_rate": 1.2985017271364141e-08, + "loss": 0.0119, + "step": 11850 + }, + { + "epoch": 19.865996649916248, + "grad_norm": 0.0661471039056778, + "learning_rate": 1.132230600141837e-08, + "loss": 0.0099, + "step": 11860 + }, + { + "epoch": 19.882747068676718, + "grad_norm": 0.05712989717721939, + "learning_rate": 9.77342377557644e-09, + "loss": 0.0116, + "step": 11870 + }, + { + "epoch": 19.899497487437188, + "grad_norm": 0.05498345196247101, + "learning_rate": 8.338382356565811e-09, + "loss": 0.0091, + "step": 11880 + }, + { + "epoch": 19.916247906197654, + "grad_norm": 0.07772324234247208, + "learning_rate": 7.017192642573833e-09, + "loss": 0.0104, + "step": 11890 + }, + { + "epoch": 19.932998324958124, + "grad_norm": 0.071397565305233, + "learning_rate": 5.809864667157827e-09, + "loss": 0.0113, + "step": 11900 + }, + { + "epoch": 19.949748743718594, + "grad_norm": 0.07486724853515625, + "learning_rate": 4.716407599170136e-09, + "loss": 0.0114, + "step": 11910 + }, + { + "epoch": 19.96649916247906, + "grad_norm": 0.06778570264577866, + "learning_rate": 3.736829742693182e-09, + "loss": 0.0115, + "step": 11920 + }, + { + "epoch": 19.98324958123953, + "grad_norm": 0.0631963312625885, + "learning_rate": 2.8711385369678543e-09, + "loss": 0.0129, + "step": 11930 + }, + { + "epoch": 20.0, + "grad_norm": 0.078452467918396, + "learning_rate": 2.1193405563418867e-09, + "loss": 0.0116, + "step": 11940 + }, + { + "epoch": 20.01675041876047, + "grad_norm": 0.05296291038393974, + "learning_rate": 1.4814415102248903e-09, + "loss": 0.0124, + "step": 11950 + }, + { + "epoch": 20.03350083752094, + "grad_norm": 0.08162922412157059, + "learning_rate": 9.574462430333998e-10, + "loss": 0.0103, + "step": 11960 + }, + { + "epoch": 20.050251256281406, + "grad_norm": 0.06841154396533966, + "learning_rate": 5.473587341642273e-10, + "loss": 0.0124, + "step": 11970 + }, + { + "epoch": 20.067001675041876, + "grad_norm": 0.07906953245401382, + "learning_rate": 2.5118209795782497e-10, + "loss": 0.0101, + "step": 11980 + }, + { + "epoch": 20.083752093802346, + "grad_norm": 0.0642547532916069, + "learning_rate": 6.891858368163195e-11, + "loss": 0.0098, + "step": 11990 + }, + { + "epoch": 20.100502512562816, + "grad_norm": 0.07142816483974457, + "learning_rate": 5.695755050938445e-13, + "loss": 0.0117, + "step": 12000 + } + ], + "logging_steps": 10, + "max_steps": 12000, + "num_input_tokens_seen": 0, + "num_train_epochs": 21, + "save_steps": 12001, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-2000/config.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-2000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..47d7b77e4255263f3803a4b6f94ca6831e459b70 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-2000/config.json @@ -0,0 +1,64 @@ +{ + "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" + ], + "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", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-2000/experiment_cfg/metadata.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-2000/experiment_cfg/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..27700cac5d292d4d42897b3c2e80bdc58e55c60e --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-2000/experiment_cfg/metadata.json @@ -0,0 +1,431 @@ +{ + "new_embodiment": { + "statistics": { + "state": { + "base_position": { + "max": [ + 7.778976917266846, + 0.33403685688972473, + 0.7223060727119446 + ], + "min": [ + -4.821934223175049, + -6.785110950469971, + 0.6994727253913879 + ], + "mean": [ + 2.5719731235192254, + -2.1489371002267235, + 0.7003612741268546 + ], + "std": [ + 1.669945884010784, + 1.3739411381361726, + 0.0009413282367754987 + ], + "q01": [ + -0.7944286339399775, + -5.486611518471242, + 0.699977555971739 + ], + "q99": [ + 6.019886905263648, + -0.5229184210026855, + 0.702930326282967 + ] + }, + "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, + 0.248543484335273, + 0.5959317405459531 + ], + "std": [ + 0.0, + 0.0, + 0.6655550240074092, + 0.374283995143559 + ], + "q01": [ + 0.0, + 0.0, + -1.0000000000000002, + 4.503254968910557e-06 + ], + "q99": [ + 0.0, + 0.0, + 1.0000000000000002, + 1.0000000000000002 + ] + }, + "end_effector_position_relative": { + "max": [ + 0.8848381638526917, + 0.8569459915161133, + 1.053884744644165 + ], + "min": [ + -0.5828549861907959, + -0.8486236333847046, + -0.3446122705936432 + ], + "mean": [ + 0.2465490874130862, + -0.004711658908481467, + 0.43463835464312073 + ], + "std": [ + 0.15798784670671595, + 0.25624102400497073, + 0.20836028727974915 + ], + "q01": [ + -0.020752051134640608, + -0.4837933887347875, + 0.1802942952289915 + ], + "q99": [ + 0.5386784714844007, + 0.4801424212024721, + 0.695095732975766 + ] + }, + "end_effector_rotation_relative": { + "max": [ + 0.9999949932098389, + 0.9997354745864868, + 0.9900747537612915, + 0.9735639095306396 + ], + "min": [ + -0.9999914169311523, + -0.9999884366989136, + -0.9975369572639465, + 6.754255821306288e-08 + ], + "mean": [ + -0.22531859714056926, + 0.046973702821026614, + -0.0568176967878771, + 0.12364307042955697 + ], + "std": [ + 0.8502144333612052, + 0.29591250318147405, + 0.30218053156837016, + 0.1636031353412931 + ], + "q01": [ + -0.9959057387172566, + -0.5665384745573048, + -0.5329334620834977, + 0.0014372279999397553 + ], + "q99": [ + 0.9904107951565839, + 0.6359623612021396, + 0.5026076022439377, + 0.42955559234879914 + ] + }, + "gripper_qpos": { + "max": [ + 0.05289580300450325, + 0.00937709677964449 + ], + "min": [ + -0.011669117026031017, + -0.06307756155729294 + ], + "mean": [ + 0.03188757060429838, + -0.0321793333086053 + ], + "std": [ + 0.01171052313737148, + 0.011427336801203189 + ], + "q01": [ + 0.006462901319459925, + -0.040557140599419166 + ], + "q99": [ + 0.04037619335388386, + -0.006523947835993347 + ] + } + }, + "action": { + "base_motion": { + "max": [ + 1.0, + 1.0, + 1.0, + 0.0 + ], + "min": [ + -1.0, + -1.0, + -1.0, + 0.0 + ], + "mean": [ + 0.006639007355857881, + 0.0027330421795561874, + -0.0017774441970214892, + 0.0 + ], + "std": [ + 0.11841528114040736, + 0.1041359083946544, + 0.08346265521208351, + 0.0 + ], + "q01": [ + -0.17562359675270237, + -0.21706402321529558, + -0.16273392585723104, + 0.0 + ], + "q99": [ + 0.2572918701644887, + 0.27022325575787115, + 0.1369626751049278, + 0.0 + ] + }, + "control_mode": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.9205539608927732 + ], + "std": [ + 0.39063657242534666 + ], + "q01": [ + -1.0 + ], + "q99": [ + -0.17711863866562935 + ] + }, + "end_effector_position": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + -0.00924803964621179, + 0.004053841761390575, + -0.08385115578460103 + ], + "std": [ + 0.45413072518223824, + 0.40158612714634856, + 0.41068634292849304 + ], + "q01": [ + -0.9669817145875343, + -0.9241307799783718, + -0.8264079503068825 + ], + "q99": [ + 0.8021511975722166, + 0.9241346582748478, + 0.8348116781502298 + ] + }, + "end_effector_rotation": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.007644157504353809, + -0.019962543093886335, + -0.002047110452602617 + ], + "std": [ + 0.11076012154791087, + 0.13176735140285795, + 0.12286528665895614 + ], + "q01": [ + -0.2659105439401192, + -0.3961694033813533, + -0.3434870832468936 + ], + "q99": [ + 0.3145786959266369, + 0.31309430361227547, + 0.34107754401942986 + ] + }, + "gripper_close": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.29870612267533575 + ], + "std": [ + 0.9543487555375691 + ], + "q01": [ + -1.0 + ], + "q99": [ + 0.7610674802484976 + ] + } + } + }, + "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/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-2000/model.safetensors.index.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-2000/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..5ef5f247b61eb93e89601c7da8334fc087e19750 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-2000/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/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-2000/trainer_state.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-2000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..c3d17c411dbd71d78547f195fab37753de3ec3ee --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-2000/trainer_state.json @@ -0,0 +1,1434 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 3.3500837520938025, + "eval_steps": 500, + "global_step": 2000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.01675041876046901, + "grad_norm": 1.1124801635742188, + "learning_rate": 4.5e-07, + "loss": 0.2576, + "step": 10 + }, + { + "epoch": 0.03350083752093802, + "grad_norm": 1.3764758110046387, + "learning_rate": 9.500000000000001e-07, + "loss": 0.2541, + "step": 20 + }, + { + "epoch": 0.05025125628140704, + "grad_norm": 1.0254155397415161, + "learning_rate": 1.45e-06, + "loss": 0.2369, + "step": 30 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 0.6008508801460266, + "learning_rate": 1.95e-06, + "loss": 0.2272, + "step": 40 + }, + { + "epoch": 0.08375209380234507, + "grad_norm": 1.085799217224121, + "learning_rate": 2.45e-06, + "loss": 0.2211, + "step": 50 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 0.9870124459266663, + "learning_rate": 2.9499999999999997e-06, + "loss": 0.2511, + "step": 60 + }, + { + "epoch": 0.11725293132328309, + "grad_norm": 0.9730237722396851, + "learning_rate": 3.4500000000000004e-06, + "loss": 0.2414, + "step": 70 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 0.792273998260498, + "learning_rate": 3.9499999999999995e-06, + "loss": 0.2311, + "step": 80 + }, + { + "epoch": 0.1507537688442211, + "grad_norm": 1.078325629234314, + "learning_rate": 4.450000000000001e-06, + "loss": 0.2185, + "step": 90 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 0.6237806081771851, + "learning_rate": 4.95e-06, + "loss": 0.1754, + "step": 100 + }, + { + "epoch": 0.18425460636515914, + "grad_norm": 0.7615986466407776, + "learning_rate": 5.45e-06, + "loss": 0.167, + "step": 110 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 0.6572969555854797, + "learning_rate": 5.95e-06, + "loss": 0.1568, + "step": 120 + }, + { + "epoch": 0.21775544388609716, + "grad_norm": 0.5472678542137146, + "learning_rate": 6.45e-06, + "loss": 0.1421, + "step": 130 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 0.4665081202983856, + "learning_rate": 6.95e-06, + "loss": 0.1526, + "step": 140 + }, + { + "epoch": 0.25125628140703515, + "grad_norm": 0.5054625272750854, + "learning_rate": 7.45e-06, + "loss": 0.1246, + "step": 150 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 0.5371837019920349, + "learning_rate": 7.95e-06, + "loss": 0.1179, + "step": 160 + }, + { + "epoch": 0.2847571189279732, + "grad_norm": 0.40661293268203735, + "learning_rate": 8.45e-06, + "loss": 0.1062, + "step": 170 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 0.2422139048576355, + "learning_rate": 8.95e-06, + "loss": 0.0963, + "step": 180 + }, + { + "epoch": 0.31825795644891125, + "grad_norm": 0.3034665584564209, + "learning_rate": 9.450000000000001e-06, + "loss": 0.0975, + "step": 190 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 0.2536943256855011, + "learning_rate": 9.95e-06, + "loss": 0.0934, + "step": 200 + }, + { + "epoch": 0.35175879396984927, + "grad_norm": 0.2668403685092926, + "learning_rate": 1.045e-05, + "loss": 0.0933, + "step": 210 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 0.22533497214317322, + "learning_rate": 1.095e-05, + "loss": 0.0792, + "step": 220 + }, + { + "epoch": 0.3852596314907873, + "grad_norm": 0.2040107697248459, + "learning_rate": 1.145e-05, + "loss": 0.0812, + "step": 230 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 0.19936227798461914, + "learning_rate": 1.1949999999999999e-05, + "loss": 0.0838, + "step": 240 + }, + { + "epoch": 0.4187604690117253, + "grad_norm": 0.2014681100845337, + "learning_rate": 1.245e-05, + "loss": 0.0804, + "step": 250 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 0.20712223649024963, + "learning_rate": 1.295e-05, + "loss": 0.0751, + "step": 260 + }, + { + "epoch": 0.45226130653266333, + "grad_norm": 0.23043803870677948, + "learning_rate": 1.345e-05, + "loss": 0.0747, + "step": 270 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 0.19819584488868713, + "learning_rate": 1.395e-05, + "loss": 0.0715, + "step": 280 + }, + { + "epoch": 0.48576214405360135, + "grad_norm": 0.2228541076183319, + "learning_rate": 1.445e-05, + "loss": 0.0708, + "step": 290 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 0.17452232539653778, + "learning_rate": 1.4950000000000001e-05, + "loss": 0.0685, + "step": 300 + }, + { + "epoch": 0.5192629815745393, + "grad_norm": 0.20126119256019592, + "learning_rate": 1.545e-05, + "loss": 0.0703, + "step": 310 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 0.2398366630077362, + "learning_rate": 1.5949999999999998e-05, + "loss": 0.0664, + "step": 320 + }, + { + "epoch": 0.5527638190954773, + "grad_norm": 0.21752075850963593, + "learning_rate": 1.645e-05, + "loss": 0.0679, + "step": 330 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 0.17013588547706604, + "learning_rate": 1.695e-05, + "loss": 0.0705, + "step": 340 + }, + { + "epoch": 0.5862646566164154, + "grad_norm": 0.2016950249671936, + "learning_rate": 1.745e-05, + "loss": 0.0687, + "step": 350 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 0.1707467883825302, + "learning_rate": 1.7950000000000003e-05, + "loss": 0.0675, + "step": 360 + }, + { + "epoch": 0.6197654941373534, + "grad_norm": 0.17180311679840088, + "learning_rate": 1.845e-05, + "loss": 0.0618, + "step": 370 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 0.21700109541416168, + "learning_rate": 1.8950000000000003e-05, + "loss": 0.0631, + "step": 380 + }, + { + "epoch": 0.6532663316582915, + "grad_norm": 0.21110156178474426, + "learning_rate": 1.945e-05, + "loss": 0.061, + "step": 390 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 0.17809593677520752, + "learning_rate": 1.995e-05, + "loss": 0.0585, + "step": 400 + }, + { + "epoch": 0.6867671691792295, + "grad_norm": 0.15697689354419708, + "learning_rate": 2.045e-05, + "loss": 0.0615, + "step": 410 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 0.17097905278205872, + "learning_rate": 2.095e-05, + "loss": 0.0568, + "step": 420 + }, + { + "epoch": 0.7202680067001676, + "grad_norm": 0.17003680765628815, + "learning_rate": 2.145e-05, + "loss": 0.0609, + "step": 430 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 0.20018595457077026, + "learning_rate": 2.195e-05, + "loss": 0.0538, + "step": 440 + }, + { + "epoch": 0.7537688442211056, + "grad_norm": 0.1872512400150299, + "learning_rate": 2.245e-05, + "loss": 0.0562, + "step": 450 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 0.210875004529953, + "learning_rate": 2.2950000000000002e-05, + "loss": 0.0538, + "step": 460 + }, + { + "epoch": 0.7872696817420436, + "grad_norm": 0.19456243515014648, + "learning_rate": 2.345e-05, + "loss": 0.0548, + "step": 470 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 0.19903069734573364, + "learning_rate": 2.395e-05, + "loss": 0.0536, + "step": 480 + }, + { + "epoch": 0.8207705192629816, + "grad_norm": 0.1998818963766098, + "learning_rate": 2.4449999999999998e-05, + "loss": 0.0533, + "step": 490 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 0.22372868657112122, + "learning_rate": 2.495e-05, + "loss": 0.0532, + "step": 500 + }, + { + "epoch": 0.8542713567839196, + "grad_norm": 0.2313748300075531, + "learning_rate": 2.5450000000000002e-05, + "loss": 0.049, + "step": 510 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 0.1595960259437561, + "learning_rate": 2.595e-05, + "loss": 0.0513, + "step": 520 + }, + { + "epoch": 0.8877721943048577, + "grad_norm": 0.3054298162460327, + "learning_rate": 2.6450000000000003e-05, + "loss": 0.0489, + "step": 530 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 0.15157485008239746, + "learning_rate": 2.695e-05, + "loss": 0.0501, + "step": 540 + }, + { + "epoch": 0.9212730318257957, + "grad_norm": 0.19291409850120544, + "learning_rate": 2.7450000000000003e-05, + "loss": 0.0428, + "step": 550 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 0.22616443037986755, + "learning_rate": 2.795e-05, + "loss": 0.048, + "step": 560 + }, + { + "epoch": 0.9547738693467337, + "grad_norm": 0.21530359983444214, + "learning_rate": 2.845e-05, + "loss": 0.051, + "step": 570 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 0.3027316629886627, + "learning_rate": 2.895e-05, + "loss": 0.0462, + "step": 580 + }, + { + "epoch": 0.9882747068676717, + "grad_norm": 0.16115690767765045, + "learning_rate": 2.945e-05, + "loss": 0.0438, + "step": 590 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 0.20157869160175323, + "learning_rate": 2.995e-05, + "loss": 0.0451, + "step": 600 + }, + { + "epoch": 1.0217755443886096, + "grad_norm": 0.19205620884895325, + "learning_rate": 2.9999953864407506e-05, + "loss": 0.0437, + "step": 610 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 0.1747637391090393, + "learning_rate": 2.99997943837114e-05, + "loss": 0.0454, + "step": 620 + }, + { + "epoch": 1.0552763819095476, + "grad_norm": 0.18132591247558594, + "learning_rate": 2.9999520989547326e-05, + "loss": 0.0488, + "step": 630 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 0.1378272920846939, + "learning_rate": 2.9999133683991525e-05, + "loss": 0.042, + "step": 640 + }, + { + "epoch": 1.0887772194304857, + "grad_norm": 0.17927886545658112, + "learning_rate": 2.9998632469985326e-05, + "loss": 0.0452, + "step": 650 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 0.1842069774866104, + "learning_rate": 2.999801735133512e-05, + "loss": 0.0418, + "step": 660 + }, + { + "epoch": 1.1222780569514237, + "grad_norm": 0.1894877851009369, + "learning_rate": 2.9997288332712325e-05, + "loss": 0.0421, + "step": 670 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 0.17502732574939728, + "learning_rate": 2.999644541965335e-05, + "loss": 0.0395, + "step": 680 + }, + { + "epoch": 1.1557788944723617, + "grad_norm": 0.19280853867530823, + "learning_rate": 2.9995488618559554e-05, + "loss": 0.0383, + "step": 690 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 0.21065080165863037, + "learning_rate": 2.9994417936697218e-05, + "loss": 0.0409, + "step": 700 + }, + { + "epoch": 1.1892797319932997, + "grad_norm": 0.15386411547660828, + "learning_rate": 2.9993233382197454e-05, + "loss": 0.0417, + "step": 710 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 0.18249723315238953, + "learning_rate": 2.999193496405616e-05, + "loss": 0.0386, + "step": 720 + }, + { + "epoch": 1.2227805695142377, + "grad_norm": 0.17197810113430023, + "learning_rate": 2.9990522692133962e-05, + "loss": 0.038, + "step": 730 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 0.15405157208442688, + "learning_rate": 2.9988996577156134e-05, + "loss": 0.0421, + "step": 740 + }, + { + "epoch": 1.2562814070351758, + "grad_norm": 0.16718165576457977, + "learning_rate": 2.9987356630712496e-05, + "loss": 0.04, + "step": 750 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 0.1574426144361496, + "learning_rate": 2.9985602865257354e-05, + "loss": 0.0403, + "step": 760 + }, + { + "epoch": 1.2897822445561138, + "grad_norm": 0.11435810476541519, + "learning_rate": 2.998373529410939e-05, + "loss": 0.0398, + "step": 770 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 0.16009581089019775, + "learning_rate": 2.9981753931451567e-05, + "loss": 0.037, + "step": 780 + }, + { + "epoch": 1.3232830820770518, + "grad_norm": 0.15292680263519287, + "learning_rate": 2.9979658792331016e-05, + "loss": 0.0375, + "step": 790 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 0.1592884510755539, + "learning_rate": 2.997744989265893e-05, + "loss": 0.037, + "step": 800 + }, + { + "epoch": 1.3567839195979898, + "grad_norm": 0.22414714097976685, + "learning_rate": 2.9975127249210426e-05, + "loss": 0.0344, + "step": 810 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 0.14030888676643372, + "learning_rate": 2.9972690879624445e-05, + "loss": 0.0364, + "step": 820 + }, + { + "epoch": 1.3902847571189278, + "grad_norm": 0.15426629781723022, + "learning_rate": 2.9970140802403587e-05, + "loss": 0.0368, + "step": 830 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 0.1377844214439392, + "learning_rate": 2.9967477036914002e-05, + "loss": 0.0365, + "step": 840 + }, + { + "epoch": 1.4237855946398659, + "grad_norm": 0.1314632147550583, + "learning_rate": 2.996469960338521e-05, + "loss": 0.0364, + "step": 850 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 0.1743793785572052, + "learning_rate": 2.9961808522909976e-05, + "loss": 0.0383, + "step": 860 + }, + { + "epoch": 1.457286432160804, + "grad_norm": 0.1816023737192154, + "learning_rate": 2.995880381744414e-05, + "loss": 0.0357, + "step": 870 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 0.19835101068019867, + "learning_rate": 2.995568550980643e-05, + "loss": 0.0383, + "step": 880 + }, + { + "epoch": 1.490787269681742, + "grad_norm": 0.13531947135925293, + "learning_rate": 2.9952453623678334e-05, + "loss": 0.0338, + "step": 890 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 0.13880357146263123, + "learning_rate": 2.9949108183603867e-05, + "loss": 0.036, + "step": 900 + }, + { + "epoch": 1.52428810720268, + "grad_norm": 0.18743862211704254, + "learning_rate": 2.994564921498943e-05, + "loss": 0.0358, + "step": 910 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 0.1355157345533371, + "learning_rate": 2.9942076744103594e-05, + "loss": 0.0342, + "step": 920 + }, + { + "epoch": 1.557788944723618, + "grad_norm": 0.16391365230083466, + "learning_rate": 2.9938390798076886e-05, + "loss": 0.0359, + "step": 930 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 0.1717372089624405, + "learning_rate": 2.993459140490163e-05, + "loss": 0.0344, + "step": 940 + }, + { + "epoch": 1.591289782244556, + "grad_norm": 0.19507038593292236, + "learning_rate": 2.9930678593431684e-05, + "loss": 0.0325, + "step": 950 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 0.14840291440486908, + "learning_rate": 2.9926652393382258e-05, + "loss": 0.0334, + "step": 960 + }, + { + "epoch": 1.624790619765494, + "grad_norm": 0.15482094883918762, + "learning_rate": 2.9922512835329655e-05, + "loss": 0.0338, + "step": 970 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 0.12408526986837387, + "learning_rate": 2.9918259950711073e-05, + "loss": 0.0324, + "step": 980 + }, + { + "epoch": 1.658291457286432, + "grad_norm": 0.1402246356010437, + "learning_rate": 2.991389377182435e-05, + "loss": 0.0363, + "step": 990 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 0.15740147233009338, + "learning_rate": 2.99094143318277e-05, + "loss": 0.0328, + "step": 1000 + }, + { + "epoch": 1.69179229480737, + "grad_norm": 0.14837925136089325, + "learning_rate": 2.9904821664739493e-05, + "loss": 0.0332, + "step": 1010 + }, + { + "epoch": 1.708542713567839, + "grad_norm": 0.1531188040971756, + "learning_rate": 2.990011580543799e-05, + "loss": 0.0326, + "step": 1020 + }, + { + "epoch": 1.725293132328308, + "grad_norm": 0.1466841697692871, + "learning_rate": 2.9895296789661065e-05, + "loss": 0.0326, + "step": 1030 + }, + { + "epoch": 1.742043551088777, + "grad_norm": 0.14380747079849243, + "learning_rate": 2.989036465400594e-05, + "loss": 0.0311, + "step": 1040 + }, + { + "epoch": 1.758793969849246, + "grad_norm": 0.15117715299129486, + "learning_rate": 2.9885319435928896e-05, + "loss": 0.0308, + "step": 1050 + }, + { + "epoch": 1.775544388609715, + "grad_norm": 0.1409190148115158, + "learning_rate": 2.9880161173745017e-05, + "loss": 0.0288, + "step": 1060 + }, + { + "epoch": 1.792294807370184, + "grad_norm": 0.11704489588737488, + "learning_rate": 2.9874889906627882e-05, + "loss": 0.0314, + "step": 1070 + }, + { + "epoch": 1.809045226130653, + "grad_norm": 0.1579095870256424, + "learning_rate": 2.9869505674609253e-05, + "loss": 0.0334, + "step": 1080 + }, + { + "epoch": 1.8257956448911221, + "grad_norm": 0.14541688561439514, + "learning_rate": 2.9864008518578794e-05, + "loss": 0.0301, + "step": 1090 + }, + { + "epoch": 1.8425460636515911, + "grad_norm": 0.11781132966279984, + "learning_rate": 2.9858398480283758e-05, + "loss": 0.031, + "step": 1100 + }, + { + "epoch": 1.8592964824120601, + "grad_norm": 0.1376485973596573, + "learning_rate": 2.9852675602328644e-05, + "loss": 0.0296, + "step": 1110 + }, + { + "epoch": 1.8760469011725294, + "grad_norm": 0.12675292789936066, + "learning_rate": 2.9846839928174923e-05, + "loss": 0.0289, + "step": 1120 + }, + { + "epoch": 1.8927973199329984, + "grad_norm": 0.14450091123580933, + "learning_rate": 2.984089150214065e-05, + "loss": 0.0297, + "step": 1130 + }, + { + "epoch": 1.9095477386934674, + "grad_norm": 0.15321670472621918, + "learning_rate": 2.9834830369400165e-05, + "loss": 0.0295, + "step": 1140 + }, + { + "epoch": 1.9262981574539364, + "grad_norm": 0.13381171226501465, + "learning_rate": 2.9828656575983747e-05, + "loss": 0.0279, + "step": 1150 + }, + { + "epoch": 1.9430485762144054, + "grad_norm": 0.20773489773273468, + "learning_rate": 2.9822370168777253e-05, + "loss": 0.0305, + "step": 1160 + }, + { + "epoch": 1.9597989949748744, + "grad_norm": 0.1649148315191269, + "learning_rate": 2.981597119552177e-05, + "loss": 0.0318, + "step": 1170 + }, + { + "epoch": 1.9765494137353434, + "grad_norm": 0.15318723022937775, + "learning_rate": 2.9809459704813236e-05, + "loss": 0.0283, + "step": 1180 + }, + { + "epoch": 1.9932998324958124, + "grad_norm": 0.11235921084880829, + "learning_rate": 2.980283574610211e-05, + "loss": 0.0281, + "step": 1190 + }, + { + "epoch": 2.0100502512562812, + "grad_norm": 0.14214928448200226, + "learning_rate": 2.9796099369692946e-05, + "loss": 0.0277, + "step": 1200 + }, + { + "epoch": 2.0268006700167502, + "grad_norm": 0.1358565390110016, + "learning_rate": 2.978925062674404e-05, + "loss": 0.0281, + "step": 1210 + }, + { + "epoch": 2.0435510887772192, + "grad_norm": 0.1451619565486908, + "learning_rate": 2.9782289569267057e-05, + "loss": 0.0279, + "step": 1220 + }, + { + "epoch": 2.0603015075376883, + "grad_norm": 0.1291060596704483, + "learning_rate": 2.977521625012659e-05, + "loss": 0.0307, + "step": 1230 + }, + { + "epoch": 2.0770519262981573, + "grad_norm": 0.13382935523986816, + "learning_rate": 2.9768030723039798e-05, + "loss": 0.0281, + "step": 1240 + }, + { + "epoch": 2.0938023450586263, + "grad_norm": 0.1389288753271103, + "learning_rate": 2.9760733042575984e-05, + "loss": 0.0286, + "step": 1250 + }, + { + "epoch": 2.1105527638190953, + "grad_norm": 0.13549239933490753, + "learning_rate": 2.9753323264156184e-05, + "loss": 0.0283, + "step": 1260 + }, + { + "epoch": 2.1273031825795643, + "grad_norm": 0.13530868291854858, + "learning_rate": 2.974580144405273e-05, + "loss": 0.0281, + "step": 1270 + }, + { + "epoch": 2.1440536013400333, + "grad_norm": 0.12069135159254074, + "learning_rate": 2.973816763938886e-05, + "loss": 0.0308, + "step": 1280 + }, + { + "epoch": 2.1608040201005023, + "grad_norm": 0.15732339024543762, + "learning_rate": 2.9730421908138236e-05, + "loss": 0.0274, + "step": 1290 + }, + { + "epoch": 2.1775544388609713, + "grad_norm": 0.14720749855041504, + "learning_rate": 2.9722564309124546e-05, + "loss": 0.0284, + "step": 1300 + }, + { + "epoch": 2.1943048576214403, + "grad_norm": 0.13191711902618408, + "learning_rate": 2.971459490202102e-05, + "loss": 0.0259, + "step": 1310 + }, + { + "epoch": 2.2110552763819094, + "grad_norm": 0.12703394889831543, + "learning_rate": 2.9706513747350017e-05, + "loss": 0.0245, + "step": 1320 + }, + { + "epoch": 2.2278056951423784, + "grad_norm": 0.13217298686504364, + "learning_rate": 2.9698320906482535e-05, + "loss": 0.0307, + "step": 1330 + }, + { + "epoch": 2.2445561139028474, + "grad_norm": 0.12877672910690308, + "learning_rate": 2.9690016441637747e-05, + "loss": 0.0276, + "step": 1340 + }, + { + "epoch": 2.2613065326633164, + "grad_norm": 0.13202929496765137, + "learning_rate": 2.9681600415882556e-05, + "loss": 0.0289, + "step": 1350 + }, + { + "epoch": 2.2780569514237854, + "grad_norm": 0.11203215271234512, + "learning_rate": 2.9673072893131068e-05, + "loss": 0.0274, + "step": 1360 + }, + { + "epoch": 2.2948073701842544, + "grad_norm": 0.09884139150381088, + "learning_rate": 2.9664433938144162e-05, + "loss": 0.0275, + "step": 1370 + }, + { + "epoch": 2.3115577889447234, + "grad_norm": 0.12471596896648407, + "learning_rate": 2.9655683616528958e-05, + "loss": 0.0252, + "step": 1380 + }, + { + "epoch": 2.3283082077051924, + "grad_norm": 0.15619947016239166, + "learning_rate": 2.9646821994738322e-05, + "loss": 0.0267, + "step": 1390 + }, + { + "epoch": 2.3450586264656614, + "grad_norm": 0.11800948530435562, + "learning_rate": 2.9637849140070393e-05, + "loss": 0.025, + "step": 1400 + }, + { + "epoch": 2.3618090452261304, + "grad_norm": 0.12330026179552078, + "learning_rate": 2.9628765120668033e-05, + "loss": 0.0242, + "step": 1410 + }, + { + "epoch": 2.3785594639865995, + "grad_norm": 0.14527784287929535, + "learning_rate": 2.9619570005518336e-05, + "loss": 0.0258, + "step": 1420 + }, + { + "epoch": 2.3953098827470685, + "grad_norm": 0.1384161412715912, + "learning_rate": 2.9610263864452077e-05, + "loss": 0.0245, + "step": 1430 + }, + { + "epoch": 2.4120603015075375, + "grad_norm": 0.14128261804580688, + "learning_rate": 2.9600846768143225e-05, + "loss": 0.0243, + "step": 1440 + }, + { + "epoch": 2.4288107202680065, + "grad_norm": 0.12654323875904083, + "learning_rate": 2.959131878810836e-05, + "loss": 0.0257, + "step": 1450 + }, + { + "epoch": 2.4455611390284755, + "grad_norm": 0.19983522593975067, + "learning_rate": 2.9581679996706157e-05, + "loss": 0.0278, + "step": 1460 + }, + { + "epoch": 2.4623115577889445, + "grad_norm": 0.14568129181861877, + "learning_rate": 2.9571930467136833e-05, + "loss": 0.0256, + "step": 1470 + }, + { + "epoch": 2.4790619765494135, + "grad_norm": 0.12198968231678009, + "learning_rate": 2.956207027344159e-05, + "loss": 0.0257, + "step": 1480 + }, + { + "epoch": 2.4958123953098825, + "grad_norm": 0.11267124861478806, + "learning_rate": 2.9552099490502026e-05, + "loss": 0.0246, + "step": 1490 + }, + { + "epoch": 2.5125628140703515, + "grad_norm": 0.11769300699234009, + "learning_rate": 2.9542018194039634e-05, + "loss": 0.0249, + "step": 1500 + }, + { + "epoch": 2.5293132328308205, + "grad_norm": 0.11569931358098984, + "learning_rate": 2.953182646061514e-05, + "loss": 0.0278, + "step": 1510 + }, + { + "epoch": 2.5460636515912896, + "grad_norm": 0.12251350283622742, + "learning_rate": 2.9521524367627994e-05, + "loss": 0.0233, + "step": 1520 + }, + { + "epoch": 2.5628140703517586, + "grad_norm": 0.11296738684177399, + "learning_rate": 2.9511111993315743e-05, + "loss": 0.0251, + "step": 1530 + }, + { + "epoch": 2.5795644891122276, + "grad_norm": 0.11284096539020538, + "learning_rate": 2.9500589416753447e-05, + "loss": 0.0261, + "step": 1540 + }, + { + "epoch": 2.5963149078726966, + "grad_norm": 0.14394700527191162, + "learning_rate": 2.9489956717853088e-05, + "loss": 0.0272, + "step": 1550 + }, + { + "epoch": 2.6130653266331656, + "grad_norm": 0.12496906518936157, + "learning_rate": 2.9479213977362935e-05, + "loss": 0.0255, + "step": 1560 + }, + { + "epoch": 2.6298157453936346, + "grad_norm": 0.1484213024377823, + "learning_rate": 2.9468361276866973e-05, + "loss": 0.0257, + "step": 1570 + }, + { + "epoch": 2.6465661641541036, + "grad_norm": 0.13880600035190582, + "learning_rate": 2.9457398698784243e-05, + "loss": 0.0226, + "step": 1580 + }, + { + "epoch": 2.6633165829145726, + "grad_norm": 0.1260875016450882, + "learning_rate": 2.944632632636823e-05, + "loss": 0.026, + "step": 1590 + }, + { + "epoch": 2.6800670016750416, + "grad_norm": 0.14514032006263733, + "learning_rate": 2.9435144243706255e-05, + "loss": 0.0251, + "step": 1600 + }, + { + "epoch": 2.6968174204355106, + "grad_norm": 0.10520205646753311, + "learning_rate": 2.9423852535718805e-05, + "loss": 0.023, + "step": 1610 + }, + { + "epoch": 2.7135678391959797, + "grad_norm": 0.09249555319547653, + "learning_rate": 2.9412451288158883e-05, + "loss": 0.0238, + "step": 1620 + }, + { + "epoch": 2.7303182579564487, + "grad_norm": 0.08644948154687881, + "learning_rate": 2.94009405876114e-05, + "loss": 0.0242, + "step": 1630 + }, + { + "epoch": 2.7470686767169177, + "grad_norm": 0.12189530581235886, + "learning_rate": 2.9389320521492464e-05, + "loss": 0.0216, + "step": 1640 + }, + { + "epoch": 2.7638190954773867, + "grad_norm": 0.10813789814710617, + "learning_rate": 2.9377591178048765e-05, + "loss": 0.0249, + "step": 1650 + }, + { + "epoch": 2.7805695142378557, + "grad_norm": 0.10432349145412445, + "learning_rate": 2.9365752646356857e-05, + "loss": 0.0216, + "step": 1660 + }, + { + "epoch": 2.7973199329983247, + "grad_norm": 0.08932849764823914, + "learning_rate": 2.935380501632253e-05, + "loss": 0.0245, + "step": 1670 + }, + { + "epoch": 2.8140703517587937, + "grad_norm": 0.12938445806503296, + "learning_rate": 2.934174837868008e-05, + "loss": 0.0252, + "step": 1680 + }, + { + "epoch": 2.8308207705192627, + "grad_norm": 0.20361070334911346, + "learning_rate": 2.932958282499167e-05, + "loss": 0.0258, + "step": 1690 + }, + { + "epoch": 2.8475711892797317, + "grad_norm": 0.12765756249427795, + "learning_rate": 2.9317308447646585e-05, + "loss": 0.0256, + "step": 1700 + }, + { + "epoch": 2.8643216080402008, + "grad_norm": 0.13113832473754883, + "learning_rate": 2.9304925339860565e-05, + "loss": 0.0258, + "step": 1710 + }, + { + "epoch": 2.8810720268006698, + "grad_norm": 0.13574381172657013, + "learning_rate": 2.929243359567509e-05, + "loss": 0.0231, + "step": 1720 + }, + { + "epoch": 2.8978224455611388, + "grad_norm": 0.15551957488059998, + "learning_rate": 2.9279833309956637e-05, + "loss": 0.0233, + "step": 1730 + }, + { + "epoch": 2.914572864321608, + "grad_norm": 0.1148996576666832, + "learning_rate": 2.9267124578396013e-05, + "loss": 0.0215, + "step": 1740 + }, + { + "epoch": 2.931323283082077, + "grad_norm": 0.1132749691605568, + "learning_rate": 2.9254307497507584e-05, + "loss": 0.0255, + "step": 1750 + }, + { + "epoch": 2.948073701842546, + "grad_norm": 0.12405268102884293, + "learning_rate": 2.924138216462856e-05, + "loss": 0.0232, + "step": 1760 + }, + { + "epoch": 2.964824120603015, + "grad_norm": 0.13782398402690887, + "learning_rate": 2.922834867791825e-05, + "loss": 0.0257, + "step": 1770 + }, + { + "epoch": 2.981574539363484, + "grad_norm": 0.13731901347637177, + "learning_rate": 2.9215207136357322e-05, + "loss": 0.0235, + "step": 1780 + }, + { + "epoch": 2.998324958123953, + "grad_norm": 0.11207369714975357, + "learning_rate": 2.9201957639747056e-05, + "loss": 0.0254, + "step": 1790 + }, + { + "epoch": 3.0150753768844223, + "grad_norm": 0.12427914887666702, + "learning_rate": 2.9188600288708565e-05, + "loss": 0.0231, + "step": 1800 + }, + { + "epoch": 3.0318257956448913, + "grad_norm": 0.12067142128944397, + "learning_rate": 2.9175135184682053e-05, + "loss": 0.0211, + "step": 1810 + }, + { + "epoch": 3.0485762144053603, + "grad_norm": 0.1151166558265686, + "learning_rate": 2.9161562429926038e-05, + "loss": 0.0228, + "step": 1820 + }, + { + "epoch": 3.0653266331658293, + "grad_norm": 0.12743020057678223, + "learning_rate": 2.9147882127516568e-05, + "loss": 0.0227, + "step": 1830 + }, + { + "epoch": 3.0820770519262983, + "grad_norm": 0.12181981652975082, + "learning_rate": 2.913409438134644e-05, + "loss": 0.024, + "step": 1840 + }, + { + "epoch": 3.0988274706867673, + "grad_norm": 0.1195250153541565, + "learning_rate": 2.9120199296124436e-05, + "loss": 0.0251, + "step": 1850 + }, + { + "epoch": 3.1155778894472363, + "grad_norm": 0.09719762206077576, + "learning_rate": 2.9106196977374477e-05, + "loss": 0.0232, + "step": 1860 + }, + { + "epoch": 3.1323283082077054, + "grad_norm": 0.11931390315294266, + "learning_rate": 2.9092087531434878e-05, + "loss": 0.0236, + "step": 1870 + }, + { + "epoch": 3.1490787269681744, + "grad_norm": 0.10895305871963501, + "learning_rate": 2.9077871065457495e-05, + "loss": 0.023, + "step": 1880 + }, + { + "epoch": 3.1658291457286434, + "grad_norm": 0.1338292807340622, + "learning_rate": 2.9063547687406945e-05, + "loss": 0.024, + "step": 1890 + }, + { + "epoch": 3.1825795644891124, + "grad_norm": 0.11511155962944031, + "learning_rate": 2.9049117506059756e-05, + "loss": 0.0208, + "step": 1900 + }, + { + "epoch": 3.1993299832495814, + "grad_norm": 0.11553452163934708, + "learning_rate": 2.9034580631003564e-05, + "loss": 0.0232, + "step": 1910 + }, + { + "epoch": 3.2160804020100504, + "grad_norm": 0.1008048877120018, + "learning_rate": 2.9019937172636276e-05, + "loss": 0.0222, + "step": 1920 + }, + { + "epoch": 3.2328308207705194, + "grad_norm": 0.10765612125396729, + "learning_rate": 2.9005187242165228e-05, + "loss": 0.0226, + "step": 1930 + }, + { + "epoch": 3.2495812395309884, + "grad_norm": 0.11185599863529205, + "learning_rate": 2.899033095160633e-05, + "loss": 0.0213, + "step": 1940 + }, + { + "epoch": 3.2663316582914574, + "grad_norm": 0.09481639415025711, + "learning_rate": 2.8975368413783235e-05, + "loss": 0.0217, + "step": 1950 + }, + { + "epoch": 3.2830820770519265, + "grad_norm": 0.11557790637016296, + "learning_rate": 2.8960299742326473e-05, + "loss": 0.0231, + "step": 1960 + }, + { + "epoch": 3.2998324958123955, + "grad_norm": 0.13437284529209137, + "learning_rate": 2.8945125051672595e-05, + "loss": 0.0234, + "step": 1970 + }, + { + "epoch": 3.3165829145728645, + "grad_norm": 0.11012782156467438, + "learning_rate": 2.892984445706328e-05, + "loss": 0.0225, + "step": 1980 + }, + { + "epoch": 3.3333333333333335, + "grad_norm": 0.11317654699087143, + "learning_rate": 2.891445807454449e-05, + "loss": 0.024, + "step": 1990 + }, + { + "epoch": 3.3500837520938025, + "grad_norm": 0.19488593935966492, + "learning_rate": 2.8898966020965576e-05, + "loss": 0.024, + "step": 2000 + } + ], + "logging_steps": 10, + "max_steps": 12000, + "num_input_tokens_seen": 0, + "num_train_epochs": 21, + "save_steps": 12001, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-5000/config.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-5000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..47d7b77e4255263f3803a4b6f94ca6831e459b70 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-5000/config.json @@ -0,0 +1,64 @@ +{ + "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" + ], + "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", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-5000/model.safetensors.index.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-5000/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..5ef5f247b61eb93e89601c7da8334fc087e19750 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-5000/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/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-5000/trainer_state.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-5000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..5550acef526777ecadaf56909dfcf2b98ac35d64 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-5000/trainer_state.json @@ -0,0 +1,3534 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 8.375209380234505, + "eval_steps": 500, + "global_step": 5000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.01675041876046901, + "grad_norm": 1.1124801635742188, + "learning_rate": 4.5e-07, + "loss": 0.2576, + "step": 10 + }, + { + "epoch": 0.03350083752093802, + "grad_norm": 1.3764758110046387, + "learning_rate": 9.500000000000001e-07, + "loss": 0.2541, + "step": 20 + }, + { + "epoch": 0.05025125628140704, + "grad_norm": 1.0254155397415161, + "learning_rate": 1.45e-06, + "loss": 0.2369, + "step": 30 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 0.6008508801460266, + "learning_rate": 1.95e-06, + "loss": 0.2272, + "step": 40 + }, + { + "epoch": 0.08375209380234507, + "grad_norm": 1.085799217224121, + "learning_rate": 2.45e-06, + "loss": 0.2211, + "step": 50 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 0.9870124459266663, + "learning_rate": 2.9499999999999997e-06, + "loss": 0.2511, + "step": 60 + }, + { + "epoch": 0.11725293132328309, + "grad_norm": 0.9730237722396851, + "learning_rate": 3.4500000000000004e-06, + "loss": 0.2414, + "step": 70 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 0.792273998260498, + "learning_rate": 3.9499999999999995e-06, + "loss": 0.2311, + "step": 80 + }, + { + "epoch": 0.1507537688442211, + "grad_norm": 1.078325629234314, + "learning_rate": 4.450000000000001e-06, + "loss": 0.2185, + "step": 90 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 0.6237806081771851, + "learning_rate": 4.95e-06, + "loss": 0.1754, + "step": 100 + }, + { + "epoch": 0.18425460636515914, + "grad_norm": 0.7615986466407776, + "learning_rate": 5.45e-06, + "loss": 0.167, + "step": 110 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 0.6572969555854797, + "learning_rate": 5.95e-06, + "loss": 0.1568, + "step": 120 + }, + { + "epoch": 0.21775544388609716, + "grad_norm": 0.5472678542137146, + "learning_rate": 6.45e-06, + "loss": 0.1421, + "step": 130 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 0.4665081202983856, + "learning_rate": 6.95e-06, + "loss": 0.1526, + "step": 140 + }, + { + "epoch": 0.25125628140703515, + "grad_norm": 0.5054625272750854, + "learning_rate": 7.45e-06, + "loss": 0.1246, + "step": 150 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 0.5371837019920349, + "learning_rate": 7.95e-06, + "loss": 0.1179, + "step": 160 + }, + { + "epoch": 0.2847571189279732, + "grad_norm": 0.40661293268203735, + "learning_rate": 8.45e-06, + "loss": 0.1062, + "step": 170 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 0.2422139048576355, + "learning_rate": 8.95e-06, + "loss": 0.0963, + "step": 180 + }, + { + "epoch": 0.31825795644891125, + "grad_norm": 0.3034665584564209, + "learning_rate": 9.450000000000001e-06, + "loss": 0.0975, + "step": 190 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 0.2536943256855011, + "learning_rate": 9.95e-06, + "loss": 0.0934, + "step": 200 + }, + { + "epoch": 0.35175879396984927, + "grad_norm": 0.2668403685092926, + "learning_rate": 1.045e-05, + "loss": 0.0933, + "step": 210 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 0.22533497214317322, + "learning_rate": 1.095e-05, + "loss": 0.0792, + "step": 220 + }, + { + "epoch": 0.3852596314907873, + "grad_norm": 0.2040107697248459, + "learning_rate": 1.145e-05, + "loss": 0.0812, + "step": 230 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 0.19936227798461914, + "learning_rate": 1.1949999999999999e-05, + "loss": 0.0838, + "step": 240 + }, + { + "epoch": 0.4187604690117253, + "grad_norm": 0.2014681100845337, + "learning_rate": 1.245e-05, + "loss": 0.0804, + "step": 250 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 0.20712223649024963, + "learning_rate": 1.295e-05, + "loss": 0.0751, + "step": 260 + }, + { + "epoch": 0.45226130653266333, + "grad_norm": 0.23043803870677948, + "learning_rate": 1.345e-05, + "loss": 0.0747, + "step": 270 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 0.19819584488868713, + "learning_rate": 1.395e-05, + "loss": 0.0715, + "step": 280 + }, + { + "epoch": 0.48576214405360135, + "grad_norm": 0.2228541076183319, + "learning_rate": 1.445e-05, + "loss": 0.0708, + "step": 290 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 0.17452232539653778, + "learning_rate": 1.4950000000000001e-05, + "loss": 0.0685, + "step": 300 + }, + { + "epoch": 0.5192629815745393, + "grad_norm": 0.20126119256019592, + "learning_rate": 1.545e-05, + "loss": 0.0703, + "step": 310 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 0.2398366630077362, + "learning_rate": 1.5949999999999998e-05, + "loss": 0.0664, + "step": 320 + }, + { + "epoch": 0.5527638190954773, + "grad_norm": 0.21752075850963593, + "learning_rate": 1.645e-05, + "loss": 0.0679, + "step": 330 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 0.17013588547706604, + "learning_rate": 1.695e-05, + "loss": 0.0705, + "step": 340 + }, + { + "epoch": 0.5862646566164154, + "grad_norm": 0.2016950249671936, + "learning_rate": 1.745e-05, + "loss": 0.0687, + "step": 350 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 0.1707467883825302, + "learning_rate": 1.7950000000000003e-05, + "loss": 0.0675, + "step": 360 + }, + { + "epoch": 0.6197654941373534, + "grad_norm": 0.17180311679840088, + "learning_rate": 1.845e-05, + "loss": 0.0618, + "step": 370 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 0.21700109541416168, + "learning_rate": 1.8950000000000003e-05, + "loss": 0.0631, + "step": 380 + }, + { + "epoch": 0.6532663316582915, + "grad_norm": 0.21110156178474426, + "learning_rate": 1.945e-05, + "loss": 0.061, + "step": 390 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 0.17809593677520752, + "learning_rate": 1.995e-05, + "loss": 0.0585, + "step": 400 + }, + { + "epoch": 0.6867671691792295, + "grad_norm": 0.15697689354419708, + "learning_rate": 2.045e-05, + "loss": 0.0615, + "step": 410 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 0.17097905278205872, + "learning_rate": 2.095e-05, + "loss": 0.0568, + "step": 420 + }, + { + "epoch": 0.7202680067001676, + "grad_norm": 0.17003680765628815, + "learning_rate": 2.145e-05, + "loss": 0.0609, + "step": 430 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 0.20018595457077026, + "learning_rate": 2.195e-05, + "loss": 0.0538, + "step": 440 + }, + { + "epoch": 0.7537688442211056, + "grad_norm": 0.1872512400150299, + "learning_rate": 2.245e-05, + "loss": 0.0562, + "step": 450 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 0.210875004529953, + "learning_rate": 2.2950000000000002e-05, + "loss": 0.0538, + "step": 460 + }, + { + "epoch": 0.7872696817420436, + "grad_norm": 0.19456243515014648, + "learning_rate": 2.345e-05, + "loss": 0.0548, + "step": 470 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 0.19903069734573364, + "learning_rate": 2.395e-05, + "loss": 0.0536, + "step": 480 + }, + { + "epoch": 0.8207705192629816, + "grad_norm": 0.1998818963766098, + "learning_rate": 2.4449999999999998e-05, + "loss": 0.0533, + "step": 490 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 0.22372868657112122, + "learning_rate": 2.495e-05, + "loss": 0.0532, + "step": 500 + }, + { + "epoch": 0.8542713567839196, + "grad_norm": 0.2313748300075531, + "learning_rate": 2.5450000000000002e-05, + "loss": 0.049, + "step": 510 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 0.1595960259437561, + "learning_rate": 2.595e-05, + "loss": 0.0513, + "step": 520 + }, + { + "epoch": 0.8877721943048577, + "grad_norm": 0.3054298162460327, + "learning_rate": 2.6450000000000003e-05, + "loss": 0.0489, + "step": 530 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 0.15157485008239746, + "learning_rate": 2.695e-05, + "loss": 0.0501, + "step": 540 + }, + { + "epoch": 0.9212730318257957, + "grad_norm": 0.19291409850120544, + "learning_rate": 2.7450000000000003e-05, + "loss": 0.0428, + "step": 550 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 0.22616443037986755, + "learning_rate": 2.795e-05, + "loss": 0.048, + "step": 560 + }, + { + "epoch": 0.9547738693467337, + "grad_norm": 0.21530359983444214, + "learning_rate": 2.845e-05, + "loss": 0.051, + "step": 570 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 0.3027316629886627, + "learning_rate": 2.895e-05, + "loss": 0.0462, + "step": 580 + }, + { + "epoch": 0.9882747068676717, + "grad_norm": 0.16115690767765045, + "learning_rate": 2.945e-05, + "loss": 0.0438, + "step": 590 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 0.20157869160175323, + "learning_rate": 2.995e-05, + "loss": 0.0451, + "step": 600 + }, + { + "epoch": 1.0217755443886096, + "grad_norm": 0.19205620884895325, + "learning_rate": 2.9999953864407506e-05, + "loss": 0.0437, + "step": 610 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 0.1747637391090393, + "learning_rate": 2.99997943837114e-05, + "loss": 0.0454, + "step": 620 + }, + { + "epoch": 1.0552763819095476, + "grad_norm": 0.18132591247558594, + "learning_rate": 2.9999520989547326e-05, + "loss": 0.0488, + "step": 630 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 0.1378272920846939, + "learning_rate": 2.9999133683991525e-05, + "loss": 0.042, + "step": 640 + }, + { + "epoch": 1.0887772194304857, + "grad_norm": 0.17927886545658112, + "learning_rate": 2.9998632469985326e-05, + "loss": 0.0452, + "step": 650 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 0.1842069774866104, + "learning_rate": 2.999801735133512e-05, + "loss": 0.0418, + "step": 660 + }, + { + "epoch": 1.1222780569514237, + "grad_norm": 0.1894877851009369, + "learning_rate": 2.9997288332712325e-05, + "loss": 0.0421, + "step": 670 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 0.17502732574939728, + "learning_rate": 2.999644541965335e-05, + "loss": 0.0395, + "step": 680 + }, + { + "epoch": 1.1557788944723617, + "grad_norm": 0.19280853867530823, + "learning_rate": 2.9995488618559554e-05, + "loss": 0.0383, + "step": 690 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 0.21065080165863037, + "learning_rate": 2.9994417936697218e-05, + "loss": 0.0409, + "step": 700 + }, + { + "epoch": 1.1892797319932997, + "grad_norm": 0.15386411547660828, + "learning_rate": 2.9993233382197454e-05, + "loss": 0.0417, + "step": 710 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 0.18249723315238953, + "learning_rate": 2.999193496405616e-05, + "loss": 0.0386, + "step": 720 + }, + { + "epoch": 1.2227805695142377, + "grad_norm": 0.17197810113430023, + "learning_rate": 2.9990522692133962e-05, + "loss": 0.038, + "step": 730 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 0.15405157208442688, + "learning_rate": 2.9988996577156134e-05, + "loss": 0.0421, + "step": 740 + }, + { + "epoch": 1.2562814070351758, + "grad_norm": 0.16718165576457977, + "learning_rate": 2.9987356630712496e-05, + "loss": 0.04, + "step": 750 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 0.1574426144361496, + "learning_rate": 2.9985602865257354e-05, + "loss": 0.0403, + "step": 760 + }, + { + "epoch": 1.2897822445561138, + "grad_norm": 0.11435810476541519, + "learning_rate": 2.998373529410939e-05, + "loss": 0.0398, + "step": 770 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 0.16009581089019775, + "learning_rate": 2.9981753931451567e-05, + "loss": 0.037, + "step": 780 + }, + { + "epoch": 1.3232830820770518, + "grad_norm": 0.15292680263519287, + "learning_rate": 2.9979658792331016e-05, + "loss": 0.0375, + "step": 790 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 0.1592884510755539, + "learning_rate": 2.997744989265893e-05, + "loss": 0.037, + "step": 800 + }, + { + "epoch": 1.3567839195979898, + "grad_norm": 0.22414714097976685, + "learning_rate": 2.9975127249210426e-05, + "loss": 0.0344, + "step": 810 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 0.14030888676643372, + "learning_rate": 2.9972690879624445e-05, + "loss": 0.0364, + "step": 820 + }, + { + "epoch": 1.3902847571189278, + "grad_norm": 0.15426629781723022, + "learning_rate": 2.9970140802403587e-05, + "loss": 0.0368, + "step": 830 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 0.1377844214439392, + "learning_rate": 2.9967477036914002e-05, + "loss": 0.0365, + "step": 840 + }, + { + "epoch": 1.4237855946398659, + "grad_norm": 0.1314632147550583, + "learning_rate": 2.996469960338521e-05, + "loss": 0.0364, + "step": 850 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 0.1743793785572052, + "learning_rate": 2.9961808522909976e-05, + "loss": 0.0383, + "step": 860 + }, + { + "epoch": 1.457286432160804, + "grad_norm": 0.1816023737192154, + "learning_rate": 2.995880381744414e-05, + "loss": 0.0357, + "step": 870 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 0.19835101068019867, + "learning_rate": 2.995568550980643e-05, + "loss": 0.0383, + "step": 880 + }, + { + "epoch": 1.490787269681742, + "grad_norm": 0.13531947135925293, + "learning_rate": 2.9952453623678334e-05, + "loss": 0.0338, + "step": 890 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 0.13880357146263123, + "learning_rate": 2.9949108183603867e-05, + "loss": 0.036, + "step": 900 + }, + { + "epoch": 1.52428810720268, + "grad_norm": 0.18743862211704254, + "learning_rate": 2.994564921498943e-05, + "loss": 0.0358, + "step": 910 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 0.1355157345533371, + "learning_rate": 2.9942076744103594e-05, + "loss": 0.0342, + "step": 920 + }, + { + "epoch": 1.557788944723618, + "grad_norm": 0.16391365230083466, + "learning_rate": 2.9938390798076886e-05, + "loss": 0.0359, + "step": 930 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 0.1717372089624405, + "learning_rate": 2.993459140490163e-05, + "loss": 0.0344, + "step": 940 + }, + { + "epoch": 1.591289782244556, + "grad_norm": 0.19507038593292236, + "learning_rate": 2.9930678593431684e-05, + "loss": 0.0325, + "step": 950 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 0.14840291440486908, + "learning_rate": 2.9926652393382258e-05, + "loss": 0.0334, + "step": 960 + }, + { + "epoch": 1.624790619765494, + "grad_norm": 0.15482094883918762, + "learning_rate": 2.9922512835329655e-05, + "loss": 0.0338, + "step": 970 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 0.12408526986837387, + "learning_rate": 2.9918259950711073e-05, + "loss": 0.0324, + "step": 980 + }, + { + "epoch": 1.658291457286432, + "grad_norm": 0.1402246356010437, + "learning_rate": 2.991389377182435e-05, + "loss": 0.0363, + "step": 990 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 0.15740147233009338, + "learning_rate": 2.99094143318277e-05, + "loss": 0.0328, + "step": 1000 + }, + { + "epoch": 1.69179229480737, + "grad_norm": 0.14837925136089325, + "learning_rate": 2.9904821664739493e-05, + "loss": 0.0332, + "step": 1010 + }, + { + "epoch": 1.708542713567839, + "grad_norm": 0.1531188040971756, + "learning_rate": 2.990011580543799e-05, + "loss": 0.0326, + "step": 1020 + }, + { + "epoch": 1.725293132328308, + "grad_norm": 0.1466841697692871, + "learning_rate": 2.9895296789661065e-05, + "loss": 0.0326, + "step": 1030 + }, + { + "epoch": 1.742043551088777, + "grad_norm": 0.14380747079849243, + "learning_rate": 2.989036465400594e-05, + "loss": 0.0311, + "step": 1040 + }, + { + "epoch": 1.758793969849246, + "grad_norm": 0.15117715299129486, + "learning_rate": 2.9885319435928896e-05, + "loss": 0.0308, + "step": 1050 + }, + { + "epoch": 1.775544388609715, + "grad_norm": 0.1409190148115158, + "learning_rate": 2.9880161173745017e-05, + "loss": 0.0288, + "step": 1060 + }, + { + "epoch": 1.792294807370184, + "grad_norm": 0.11704489588737488, + "learning_rate": 2.9874889906627882e-05, + "loss": 0.0314, + "step": 1070 + }, + { + "epoch": 1.809045226130653, + "grad_norm": 0.1579095870256424, + "learning_rate": 2.9869505674609253e-05, + "loss": 0.0334, + "step": 1080 + }, + { + "epoch": 1.8257956448911221, + "grad_norm": 0.14541688561439514, + "learning_rate": 2.9864008518578794e-05, + "loss": 0.0301, + "step": 1090 + }, + { + "epoch": 1.8425460636515911, + "grad_norm": 0.11781132966279984, + "learning_rate": 2.9858398480283758e-05, + "loss": 0.031, + "step": 1100 + }, + { + "epoch": 1.8592964824120601, + "grad_norm": 0.1376485973596573, + "learning_rate": 2.9852675602328644e-05, + "loss": 0.0296, + "step": 1110 + }, + { + "epoch": 1.8760469011725294, + "grad_norm": 0.12675292789936066, + "learning_rate": 2.9846839928174923e-05, + "loss": 0.0289, + "step": 1120 + }, + { + "epoch": 1.8927973199329984, + "grad_norm": 0.14450091123580933, + "learning_rate": 2.984089150214065e-05, + "loss": 0.0297, + "step": 1130 + }, + { + "epoch": 1.9095477386934674, + "grad_norm": 0.15321670472621918, + "learning_rate": 2.9834830369400165e-05, + "loss": 0.0295, + "step": 1140 + }, + { + "epoch": 1.9262981574539364, + "grad_norm": 0.13381171226501465, + "learning_rate": 2.9828656575983747e-05, + "loss": 0.0279, + "step": 1150 + }, + { + "epoch": 1.9430485762144054, + "grad_norm": 0.20773489773273468, + "learning_rate": 2.9822370168777253e-05, + "loss": 0.0305, + "step": 1160 + }, + { + "epoch": 1.9597989949748744, + "grad_norm": 0.1649148315191269, + "learning_rate": 2.981597119552177e-05, + "loss": 0.0318, + "step": 1170 + }, + { + "epoch": 1.9765494137353434, + "grad_norm": 0.15318723022937775, + "learning_rate": 2.9809459704813236e-05, + "loss": 0.0283, + "step": 1180 + }, + { + "epoch": 1.9932998324958124, + "grad_norm": 0.11235921084880829, + "learning_rate": 2.980283574610211e-05, + "loss": 0.0281, + "step": 1190 + }, + { + "epoch": 2.0100502512562812, + "grad_norm": 0.14214928448200226, + "learning_rate": 2.9796099369692946e-05, + "loss": 0.0277, + "step": 1200 + }, + { + "epoch": 2.0268006700167502, + "grad_norm": 0.1358565390110016, + "learning_rate": 2.978925062674404e-05, + "loss": 0.0281, + "step": 1210 + }, + { + "epoch": 2.0435510887772192, + "grad_norm": 0.1451619565486908, + "learning_rate": 2.9782289569267057e-05, + "loss": 0.0279, + "step": 1220 + }, + { + "epoch": 2.0603015075376883, + "grad_norm": 0.1291060596704483, + "learning_rate": 2.977521625012659e-05, + "loss": 0.0307, + "step": 1230 + }, + { + "epoch": 2.0770519262981573, + "grad_norm": 0.13382935523986816, + "learning_rate": 2.9768030723039798e-05, + "loss": 0.0281, + "step": 1240 + }, + { + "epoch": 2.0938023450586263, + "grad_norm": 0.1389288753271103, + "learning_rate": 2.9760733042575984e-05, + "loss": 0.0286, + "step": 1250 + }, + { + "epoch": 2.1105527638190953, + "grad_norm": 0.13549239933490753, + "learning_rate": 2.9753323264156184e-05, + "loss": 0.0283, + "step": 1260 + }, + { + "epoch": 2.1273031825795643, + "grad_norm": 0.13530868291854858, + "learning_rate": 2.974580144405273e-05, + "loss": 0.0281, + "step": 1270 + }, + { + "epoch": 2.1440536013400333, + "grad_norm": 0.12069135159254074, + "learning_rate": 2.973816763938886e-05, + "loss": 0.0308, + "step": 1280 + }, + { + "epoch": 2.1608040201005023, + "grad_norm": 0.15732339024543762, + "learning_rate": 2.9730421908138236e-05, + "loss": 0.0274, + "step": 1290 + }, + { + "epoch": 2.1775544388609713, + "grad_norm": 0.14720749855041504, + "learning_rate": 2.9722564309124546e-05, + "loss": 0.0284, + "step": 1300 + }, + { + "epoch": 2.1943048576214403, + "grad_norm": 0.13191711902618408, + "learning_rate": 2.971459490202102e-05, + "loss": 0.0259, + "step": 1310 + }, + { + "epoch": 2.2110552763819094, + "grad_norm": 0.12703394889831543, + "learning_rate": 2.9706513747350017e-05, + "loss": 0.0245, + "step": 1320 + }, + { + "epoch": 2.2278056951423784, + "grad_norm": 0.13217298686504364, + "learning_rate": 2.9698320906482535e-05, + "loss": 0.0307, + "step": 1330 + }, + { + "epoch": 2.2445561139028474, + "grad_norm": 0.12877672910690308, + "learning_rate": 2.9690016441637747e-05, + "loss": 0.0276, + "step": 1340 + }, + { + "epoch": 2.2613065326633164, + "grad_norm": 0.13202929496765137, + "learning_rate": 2.9681600415882556e-05, + "loss": 0.0289, + "step": 1350 + }, + { + "epoch": 2.2780569514237854, + "grad_norm": 0.11203215271234512, + "learning_rate": 2.9673072893131068e-05, + "loss": 0.0274, + "step": 1360 + }, + { + "epoch": 2.2948073701842544, + "grad_norm": 0.09884139150381088, + "learning_rate": 2.9664433938144162e-05, + "loss": 0.0275, + "step": 1370 + }, + { + "epoch": 2.3115577889447234, + "grad_norm": 0.12471596896648407, + "learning_rate": 2.9655683616528958e-05, + "loss": 0.0252, + "step": 1380 + }, + { + "epoch": 2.3283082077051924, + "grad_norm": 0.15619947016239166, + "learning_rate": 2.9646821994738322e-05, + "loss": 0.0267, + "step": 1390 + }, + { + "epoch": 2.3450586264656614, + "grad_norm": 0.11800948530435562, + "learning_rate": 2.9637849140070393e-05, + "loss": 0.025, + "step": 1400 + }, + { + "epoch": 2.3618090452261304, + "grad_norm": 0.12330026179552078, + "learning_rate": 2.9628765120668033e-05, + "loss": 0.0242, + "step": 1410 + }, + { + "epoch": 2.3785594639865995, + "grad_norm": 0.14527784287929535, + "learning_rate": 2.9619570005518336e-05, + "loss": 0.0258, + "step": 1420 + }, + { + "epoch": 2.3953098827470685, + "grad_norm": 0.1384161412715912, + "learning_rate": 2.9610263864452077e-05, + "loss": 0.0245, + "step": 1430 + }, + { + "epoch": 2.4120603015075375, + "grad_norm": 0.14128261804580688, + "learning_rate": 2.9600846768143225e-05, + "loss": 0.0243, + "step": 1440 + }, + { + "epoch": 2.4288107202680065, + "grad_norm": 0.12654323875904083, + "learning_rate": 2.959131878810836e-05, + "loss": 0.0257, + "step": 1450 + }, + { + "epoch": 2.4455611390284755, + "grad_norm": 0.19983522593975067, + "learning_rate": 2.9581679996706157e-05, + "loss": 0.0278, + "step": 1460 + }, + { + "epoch": 2.4623115577889445, + "grad_norm": 0.14568129181861877, + "learning_rate": 2.9571930467136833e-05, + "loss": 0.0256, + "step": 1470 + }, + { + "epoch": 2.4790619765494135, + "grad_norm": 0.12198968231678009, + "learning_rate": 2.956207027344159e-05, + "loss": 0.0257, + "step": 1480 + }, + { + "epoch": 2.4958123953098825, + "grad_norm": 0.11267124861478806, + "learning_rate": 2.9552099490502026e-05, + "loss": 0.0246, + "step": 1490 + }, + { + "epoch": 2.5125628140703515, + "grad_norm": 0.11769300699234009, + "learning_rate": 2.9542018194039634e-05, + "loss": 0.0249, + "step": 1500 + }, + { + "epoch": 2.5293132328308205, + "grad_norm": 0.11569931358098984, + "learning_rate": 2.953182646061514e-05, + "loss": 0.0278, + "step": 1510 + }, + { + "epoch": 2.5460636515912896, + "grad_norm": 0.12251350283622742, + "learning_rate": 2.9521524367627994e-05, + "loss": 0.0233, + "step": 1520 + }, + { + "epoch": 2.5628140703517586, + "grad_norm": 0.11296738684177399, + "learning_rate": 2.9511111993315743e-05, + "loss": 0.0251, + "step": 1530 + }, + { + "epoch": 2.5795644891122276, + "grad_norm": 0.11284096539020538, + "learning_rate": 2.9500589416753447e-05, + "loss": 0.0261, + "step": 1540 + }, + { + "epoch": 2.5963149078726966, + "grad_norm": 0.14394700527191162, + "learning_rate": 2.9489956717853088e-05, + "loss": 0.0272, + "step": 1550 + }, + { + "epoch": 2.6130653266331656, + "grad_norm": 0.12496906518936157, + "learning_rate": 2.9479213977362935e-05, + "loss": 0.0255, + "step": 1560 + }, + { + "epoch": 2.6298157453936346, + "grad_norm": 0.1484213024377823, + "learning_rate": 2.9468361276866973e-05, + "loss": 0.0257, + "step": 1570 + }, + { + "epoch": 2.6465661641541036, + "grad_norm": 0.13880600035190582, + "learning_rate": 2.9457398698784243e-05, + "loss": 0.0226, + "step": 1580 + }, + { + "epoch": 2.6633165829145726, + "grad_norm": 0.1260875016450882, + "learning_rate": 2.944632632636823e-05, + "loss": 0.026, + "step": 1590 + }, + { + "epoch": 2.6800670016750416, + "grad_norm": 0.14514032006263733, + "learning_rate": 2.9435144243706255e-05, + "loss": 0.0251, + "step": 1600 + }, + { + "epoch": 2.6968174204355106, + "grad_norm": 0.10520205646753311, + "learning_rate": 2.9423852535718805e-05, + "loss": 0.023, + "step": 1610 + }, + { + "epoch": 2.7135678391959797, + "grad_norm": 0.09249555319547653, + "learning_rate": 2.9412451288158883e-05, + "loss": 0.0238, + "step": 1620 + }, + { + "epoch": 2.7303182579564487, + "grad_norm": 0.08644948154687881, + "learning_rate": 2.94009405876114e-05, + "loss": 0.0242, + "step": 1630 + }, + { + "epoch": 2.7470686767169177, + "grad_norm": 0.12189530581235886, + "learning_rate": 2.9389320521492464e-05, + "loss": 0.0216, + "step": 1640 + }, + { + "epoch": 2.7638190954773867, + "grad_norm": 0.10813789814710617, + "learning_rate": 2.9377591178048765e-05, + "loss": 0.0249, + "step": 1650 + }, + { + "epoch": 2.7805695142378557, + "grad_norm": 0.10432349145412445, + "learning_rate": 2.9365752646356857e-05, + "loss": 0.0216, + "step": 1660 + }, + { + "epoch": 2.7973199329983247, + "grad_norm": 0.08932849764823914, + "learning_rate": 2.935380501632253e-05, + "loss": 0.0245, + "step": 1670 + }, + { + "epoch": 2.8140703517587937, + "grad_norm": 0.12938445806503296, + "learning_rate": 2.934174837868008e-05, + "loss": 0.0252, + "step": 1680 + }, + { + "epoch": 2.8308207705192627, + "grad_norm": 0.20361070334911346, + "learning_rate": 2.932958282499167e-05, + "loss": 0.0258, + "step": 1690 + }, + { + "epoch": 2.8475711892797317, + "grad_norm": 0.12765756249427795, + "learning_rate": 2.9317308447646585e-05, + "loss": 0.0256, + "step": 1700 + }, + { + "epoch": 2.8643216080402008, + "grad_norm": 0.13113832473754883, + "learning_rate": 2.9304925339860565e-05, + "loss": 0.0258, + "step": 1710 + }, + { + "epoch": 2.8810720268006698, + "grad_norm": 0.13574381172657013, + "learning_rate": 2.929243359567509e-05, + "loss": 0.0231, + "step": 1720 + }, + { + "epoch": 2.8978224455611388, + "grad_norm": 0.15551957488059998, + "learning_rate": 2.9279833309956637e-05, + "loss": 0.0233, + "step": 1730 + }, + { + "epoch": 2.914572864321608, + "grad_norm": 0.1148996576666832, + "learning_rate": 2.9267124578396013e-05, + "loss": 0.0215, + "step": 1740 + }, + { + "epoch": 2.931323283082077, + "grad_norm": 0.1132749691605568, + "learning_rate": 2.9254307497507584e-05, + "loss": 0.0255, + "step": 1750 + }, + { + "epoch": 2.948073701842546, + "grad_norm": 0.12405268102884293, + "learning_rate": 2.924138216462856e-05, + "loss": 0.0232, + "step": 1760 + }, + { + "epoch": 2.964824120603015, + "grad_norm": 0.13782398402690887, + "learning_rate": 2.922834867791825e-05, + "loss": 0.0257, + "step": 1770 + }, + { + "epoch": 2.981574539363484, + "grad_norm": 0.13731901347637177, + "learning_rate": 2.9215207136357322e-05, + "loss": 0.0235, + "step": 1780 + }, + { + "epoch": 2.998324958123953, + "grad_norm": 0.11207369714975357, + "learning_rate": 2.9201957639747056e-05, + "loss": 0.0254, + "step": 1790 + }, + { + "epoch": 3.0150753768844223, + "grad_norm": 0.12427914887666702, + "learning_rate": 2.9188600288708565e-05, + "loss": 0.0231, + "step": 1800 + }, + { + "epoch": 3.0318257956448913, + "grad_norm": 0.12067142128944397, + "learning_rate": 2.9175135184682053e-05, + "loss": 0.0211, + "step": 1810 + }, + { + "epoch": 3.0485762144053603, + "grad_norm": 0.1151166558265686, + "learning_rate": 2.9161562429926038e-05, + "loss": 0.0228, + "step": 1820 + }, + { + "epoch": 3.0653266331658293, + "grad_norm": 0.12743020057678223, + "learning_rate": 2.9147882127516568e-05, + "loss": 0.0227, + "step": 1830 + }, + { + "epoch": 3.0820770519262983, + "grad_norm": 0.12181981652975082, + "learning_rate": 2.913409438134644e-05, + "loss": 0.024, + "step": 1840 + }, + { + "epoch": 3.0988274706867673, + "grad_norm": 0.1195250153541565, + "learning_rate": 2.9120199296124436e-05, + "loss": 0.0251, + "step": 1850 + }, + { + "epoch": 3.1155778894472363, + "grad_norm": 0.09719762206077576, + "learning_rate": 2.9106196977374477e-05, + "loss": 0.0232, + "step": 1860 + }, + { + "epoch": 3.1323283082077054, + "grad_norm": 0.11931390315294266, + "learning_rate": 2.9092087531434878e-05, + "loss": 0.0236, + "step": 1870 + }, + { + "epoch": 3.1490787269681744, + "grad_norm": 0.10895305871963501, + "learning_rate": 2.9077871065457495e-05, + "loss": 0.023, + "step": 1880 + }, + { + "epoch": 3.1658291457286434, + "grad_norm": 0.1338292807340622, + "learning_rate": 2.9063547687406945e-05, + "loss": 0.024, + "step": 1890 + }, + { + "epoch": 3.1825795644891124, + "grad_norm": 0.11511155962944031, + "learning_rate": 2.9049117506059756e-05, + "loss": 0.0208, + "step": 1900 + }, + { + "epoch": 3.1993299832495814, + "grad_norm": 0.11553452163934708, + "learning_rate": 2.9034580631003564e-05, + "loss": 0.0232, + "step": 1910 + }, + { + "epoch": 3.2160804020100504, + "grad_norm": 0.1008048877120018, + "learning_rate": 2.9019937172636276e-05, + "loss": 0.0222, + "step": 1920 + }, + { + "epoch": 3.2328308207705194, + "grad_norm": 0.10765612125396729, + "learning_rate": 2.9005187242165228e-05, + "loss": 0.0226, + "step": 1930 + }, + { + "epoch": 3.2495812395309884, + "grad_norm": 0.11185599863529205, + "learning_rate": 2.899033095160633e-05, + "loss": 0.0213, + "step": 1940 + }, + { + "epoch": 3.2663316582914574, + "grad_norm": 0.09481639415025711, + "learning_rate": 2.8975368413783235e-05, + "loss": 0.0217, + "step": 1950 + }, + { + "epoch": 3.2830820770519265, + "grad_norm": 0.11557790637016296, + "learning_rate": 2.8960299742326473e-05, + "loss": 0.0231, + "step": 1960 + }, + { + "epoch": 3.2998324958123955, + "grad_norm": 0.13437284529209137, + "learning_rate": 2.8945125051672595e-05, + "loss": 0.0234, + "step": 1970 + }, + { + "epoch": 3.3165829145728645, + "grad_norm": 0.11012782156467438, + "learning_rate": 2.892984445706328e-05, + "loss": 0.0225, + "step": 1980 + }, + { + "epoch": 3.3333333333333335, + "grad_norm": 0.11317654699087143, + "learning_rate": 2.891445807454449e-05, + "loss": 0.024, + "step": 1990 + }, + { + "epoch": 3.3500837520938025, + "grad_norm": 0.19488593935966492, + "learning_rate": 2.8898966020965576e-05, + "loss": 0.024, + "step": 2000 + }, + { + "epoch": 3.3668341708542715, + "grad_norm": 0.10659220814704895, + "learning_rate": 2.8883368413978386e-05, + "loss": 0.0225, + "step": 2010 + }, + { + "epoch": 3.3835845896147405, + "grad_norm": 0.09751500934362411, + "learning_rate": 2.8867665372036375e-05, + "loss": 0.0258, + "step": 2020 + }, + { + "epoch": 3.4003350083752095, + "grad_norm": 0.09632577747106552, + "learning_rate": 2.8851857014393705e-05, + "loss": 0.0222, + "step": 2030 + }, + { + "epoch": 3.4170854271356785, + "grad_norm": 0.12793496251106262, + "learning_rate": 2.8835943461104354e-05, + "loss": 0.0214, + "step": 2040 + }, + { + "epoch": 3.4338358458961475, + "grad_norm": 0.11942226439714432, + "learning_rate": 2.8819924833021176e-05, + "loss": 0.023, + "step": 2050 + }, + { + "epoch": 3.4505862646566166, + "grad_norm": 0.10179294645786285, + "learning_rate": 2.8803801251795007e-05, + "loss": 0.0212, + "step": 2060 + }, + { + "epoch": 3.4673366834170856, + "grad_norm": 0.1228783056139946, + "learning_rate": 2.8787572839873725e-05, + "loss": 0.0225, + "step": 2070 + }, + { + "epoch": 3.4840871021775546, + "grad_norm": 0.13721901178359985, + "learning_rate": 2.8771239720501333e-05, + "loss": 0.0199, + "step": 2080 + }, + { + "epoch": 3.5008375209380236, + "grad_norm": 0.1087748259305954, + "learning_rate": 2.8754802017717015e-05, + "loss": 0.0228, + "step": 2090 + }, + { + "epoch": 3.5175879396984926, + "grad_norm": 0.14415811002254486, + "learning_rate": 2.87382598563542e-05, + "loss": 0.0241, + "step": 2100 + }, + { + "epoch": 3.5343383584589616, + "grad_norm": 0.10477285832166672, + "learning_rate": 2.8721613362039607e-05, + "loss": 0.02, + "step": 2110 + }, + { + "epoch": 3.5510887772194306, + "grad_norm": 0.10786721855401993, + "learning_rate": 2.8704862661192295e-05, + "loss": 0.0212, + "step": 2120 + }, + { + "epoch": 3.5678391959798996, + "grad_norm": 0.10725842416286469, + "learning_rate": 2.86880078810227e-05, + "loss": 0.022, + "step": 2130 + }, + { + "epoch": 3.5845896147403686, + "grad_norm": 0.11172711849212646, + "learning_rate": 2.8671049149531677e-05, + "loss": 0.021, + "step": 2140 + }, + { + "epoch": 3.6013400335008376, + "grad_norm": 0.10756686329841614, + "learning_rate": 2.8653986595509524e-05, + "loss": 0.0208, + "step": 2150 + }, + { + "epoch": 3.6180904522613067, + "grad_norm": 0.08842871338129044, + "learning_rate": 2.8636820348534987e-05, + "loss": 0.0208, + "step": 2160 + }, + { + "epoch": 3.6348408710217757, + "grad_norm": 0.10979699343442917, + "learning_rate": 2.861955053897431e-05, + "loss": 0.0204, + "step": 2170 + }, + { + "epoch": 3.6515912897822447, + "grad_norm": 0.0892506092786789, + "learning_rate": 2.8602177297980213e-05, + "loss": 0.0209, + "step": 2180 + }, + { + "epoch": 3.6683417085427137, + "grad_norm": 0.10864245146512985, + "learning_rate": 2.8584700757490913e-05, + "loss": 0.0214, + "step": 2190 + }, + { + "epoch": 3.6850921273031827, + "grad_norm": 0.12509699165821075, + "learning_rate": 2.8567121050229122e-05, + "loss": 0.0236, + "step": 2200 + }, + { + "epoch": 3.7018425460636517, + "grad_norm": 0.11275574564933777, + "learning_rate": 2.8549438309701028e-05, + "loss": 0.0237, + "step": 2210 + }, + { + "epoch": 3.7185929648241207, + "grad_norm": 0.12549647688865662, + "learning_rate": 2.8531652670195292e-05, + "loss": 0.021, + "step": 2220 + }, + { + "epoch": 3.7353433835845897, + "grad_norm": 0.11705254763364792, + "learning_rate": 2.8513764266782025e-05, + "loss": 0.0224, + "step": 2230 + }, + { + "epoch": 3.7520938023450587, + "grad_norm": 0.16595874726772308, + "learning_rate": 2.849577323531176e-05, + "loss": 0.0224, + "step": 2240 + }, + { + "epoch": 3.7688442211055277, + "grad_norm": 0.10253427922725677, + "learning_rate": 2.8477679712414424e-05, + "loss": 0.0215, + "step": 2250 + }, + { + "epoch": 3.7855946398659968, + "grad_norm": 0.10289912670850754, + "learning_rate": 2.8459483835498298e-05, + "loss": 0.0237, + "step": 2260 + }, + { + "epoch": 3.8023450586264658, + "grad_norm": 0.11400792002677917, + "learning_rate": 2.844118574274896e-05, + "loss": 0.0205, + "step": 2270 + }, + { + "epoch": 3.819095477386935, + "grad_norm": 0.11081373691558838, + "learning_rate": 2.842278557312828e-05, + "loss": 0.0184, + "step": 2280 + }, + { + "epoch": 3.835845896147404, + "grad_norm": 0.13571597635746002, + "learning_rate": 2.84042834663733e-05, + "loss": 0.0207, + "step": 2290 + }, + { + "epoch": 3.852596314907873, + "grad_norm": 0.10395380854606628, + "learning_rate": 2.8385679562995235e-05, + "loss": 0.0209, + "step": 2300 + }, + { + "epoch": 3.869346733668342, + "grad_norm": 0.10459823906421661, + "learning_rate": 2.8366974004278357e-05, + "loss": 0.0242, + "step": 2310 + }, + { + "epoch": 3.886097152428811, + "grad_norm": 0.10516241937875748, + "learning_rate": 2.834816693227895e-05, + "loss": 0.0205, + "step": 2320 + }, + { + "epoch": 3.90284757118928, + "grad_norm": 0.1241426020860672, + "learning_rate": 2.8329258489824234e-05, + "loss": 0.0227, + "step": 2330 + }, + { + "epoch": 3.919597989949749, + "grad_norm": 0.10768191516399384, + "learning_rate": 2.8310248820511248e-05, + "loss": 0.0215, + "step": 2340 + }, + { + "epoch": 3.936348408710218, + "grad_norm": 0.10944578796625137, + "learning_rate": 2.829113806870581e-05, + "loss": 0.0214, + "step": 2350 + }, + { + "epoch": 3.953098827470687, + "grad_norm": 0.11351463198661804, + "learning_rate": 2.8271926379541363e-05, + "loss": 0.022, + "step": 2360 + }, + { + "epoch": 3.969849246231156, + "grad_norm": 0.12451263517141342, + "learning_rate": 2.8252613898917923e-05, + "loss": 0.0216, + "step": 2370 + }, + { + "epoch": 3.986599664991625, + "grad_norm": 0.11138986051082611, + "learning_rate": 2.8233200773500942e-05, + "loss": 0.0215, + "step": 2380 + }, + { + "epoch": 4.0033500837520934, + "grad_norm": 0.14621508121490479, + "learning_rate": 2.8213687150720204e-05, + "loss": 0.0227, + "step": 2390 + }, + { + "epoch": 4.0201005025125625, + "grad_norm": 0.12678438425064087, + "learning_rate": 2.8194073178768703e-05, + "loss": 0.0211, + "step": 2400 + }, + { + "epoch": 4.0368509212730315, + "grad_norm": 0.10121660679578781, + "learning_rate": 2.817435900660152e-05, + "loss": 0.0191, + "step": 2410 + }, + { + "epoch": 4.0536013400335005, + "grad_norm": 0.09031516313552856, + "learning_rate": 2.8154544783934688e-05, + "loss": 0.0241, + "step": 2420 + }, + { + "epoch": 4.0703517587939695, + "grad_norm": 0.12526579201221466, + "learning_rate": 2.8134630661244055e-05, + "loss": 0.0212, + "step": 2430 + }, + { + "epoch": 4.0871021775544385, + "grad_norm": 0.09737418591976166, + "learning_rate": 2.8114616789764146e-05, + "loss": 0.0196, + "step": 2440 + }, + { + "epoch": 4.1038525963149075, + "grad_norm": 0.12097261101007462, + "learning_rate": 2.8094503321487017e-05, + "loss": 0.0208, + "step": 2450 + }, + { + "epoch": 4.1206030150753765, + "grad_norm": 0.10906875133514404, + "learning_rate": 2.8074290409161084e-05, + "loss": 0.0183, + "step": 2460 + }, + { + "epoch": 4.1373534338358455, + "grad_norm": 0.12879160046577454, + "learning_rate": 2.8053978206289984e-05, + "loss": 0.0194, + "step": 2470 + }, + { + "epoch": 4.1541038525963145, + "grad_norm": 0.0820513367652893, + "learning_rate": 2.803356686713139e-05, + "loss": 0.0214, + "step": 2480 + }, + { + "epoch": 4.1708542713567835, + "grad_norm": 0.07784825563430786, + "learning_rate": 2.801305654669587e-05, + "loss": 0.0217, + "step": 2490 + }, + { + "epoch": 4.187604690117253, + "grad_norm": 0.0847000777721405, + "learning_rate": 2.799244740074566e-05, + "loss": 0.0195, + "step": 2500 + }, + { + "epoch": 4.204355108877722, + "grad_norm": 0.10731722414493561, + "learning_rate": 2.797173958579353e-05, + "loss": 0.0206, + "step": 2510 + }, + { + "epoch": 4.221105527638191, + "grad_norm": 0.1059015691280365, + "learning_rate": 2.7950933259101575e-05, + "loss": 0.0189, + "step": 2520 + }, + { + "epoch": 4.23785594639866, + "grad_norm": 0.10778709501028061, + "learning_rate": 2.793002857868001e-05, + "loss": 0.0208, + "step": 2530 + }, + { + "epoch": 4.254606365159129, + "grad_norm": 0.08753228187561035, + "learning_rate": 2.7909025703285983e-05, + "loss": 0.0195, + "step": 2540 + }, + { + "epoch": 4.271356783919598, + "grad_norm": 0.1175403892993927, + "learning_rate": 2.7887924792422385e-05, + "loss": 0.023, + "step": 2550 + }, + { + "epoch": 4.288107202680067, + "grad_norm": 0.12728635966777802, + "learning_rate": 2.78667260063366e-05, + "loss": 0.0215, + "step": 2560 + }, + { + "epoch": 4.304857621440536, + "grad_norm": 0.10965393483638763, + "learning_rate": 2.784542950601931e-05, + "loss": 0.019, + "step": 2570 + }, + { + "epoch": 4.321608040201005, + "grad_norm": 0.10030016303062439, + "learning_rate": 2.7824035453203293e-05, + "loss": 0.0214, + "step": 2580 + }, + { + "epoch": 4.338358458961474, + "grad_norm": 0.10159602761268616, + "learning_rate": 2.7802544010362154e-05, + "loss": 0.018, + "step": 2590 + }, + { + "epoch": 4.355108877721943, + "grad_norm": 0.08524280786514282, + "learning_rate": 2.778095534070912e-05, + "loss": 0.0225, + "step": 2600 + }, + { + "epoch": 4.371859296482412, + "grad_norm": 0.08864463865756989, + "learning_rate": 2.7759269608195777e-05, + "loss": 0.0214, + "step": 2610 + }, + { + "epoch": 4.388609715242881, + "grad_norm": 0.0836586281657219, + "learning_rate": 2.773748697751086e-05, + "loss": 0.019, + "step": 2620 + }, + { + "epoch": 4.40536013400335, + "grad_norm": 0.10019984096288681, + "learning_rate": 2.7715607614078956e-05, + "loss": 0.0204, + "step": 2630 + }, + { + "epoch": 4.422110552763819, + "grad_norm": 0.097859688103199, + "learning_rate": 2.7693631684059305e-05, + "loss": 0.0217, + "step": 2640 + }, + { + "epoch": 4.438860971524288, + "grad_norm": 0.09158801287412643, + "learning_rate": 2.767155935434448e-05, + "loss": 0.0197, + "step": 2650 + }, + { + "epoch": 4.455611390284757, + "grad_norm": 0.09695200622081757, + "learning_rate": 2.7649390792559146e-05, + "loss": 0.0206, + "step": 2660 + }, + { + "epoch": 4.472361809045226, + "grad_norm": 0.09822610765695572, + "learning_rate": 2.7627126167058807e-05, + "loss": 0.0203, + "step": 2670 + }, + { + "epoch": 4.489112227805695, + "grad_norm": 0.09746448695659637, + "learning_rate": 2.7604765646928482e-05, + "loss": 0.0204, + "step": 2680 + }, + { + "epoch": 4.505862646566165, + "grad_norm": 0.10244615375995636, + "learning_rate": 2.758230940198147e-05, + "loss": 0.0215, + "step": 2690 + }, + { + "epoch": 4.522613065326633, + "grad_norm": 0.10013485699892044, + "learning_rate": 2.755975760275802e-05, + "loss": 0.0198, + "step": 2700 + }, + { + "epoch": 4.539363484087103, + "grad_norm": 0.09892448037862778, + "learning_rate": 2.7537110420524057e-05, + "loss": 0.0199, + "step": 2710 + }, + { + "epoch": 4.556113902847571, + "grad_norm": 0.09825485199689865, + "learning_rate": 2.7514368027269875e-05, + "loss": 0.0186, + "step": 2720 + }, + { + "epoch": 4.572864321608041, + "grad_norm": 0.11656782776117325, + "learning_rate": 2.749153059570883e-05, + "loss": 0.0217, + "step": 2730 + }, + { + "epoch": 4.589614740368509, + "grad_norm": 0.08932335674762726, + "learning_rate": 2.7468598299276044e-05, + "loss": 0.0197, + "step": 2740 + }, + { + "epoch": 4.606365159128979, + "grad_norm": 0.09198352694511414, + "learning_rate": 2.7445571312127058e-05, + "loss": 0.0168, + "step": 2750 + }, + { + "epoch": 4.623115577889447, + "grad_norm": 0.09846565872430801, + "learning_rate": 2.7422449809136538e-05, + "loss": 0.0183, + "step": 2760 + }, + { + "epoch": 4.639865996649917, + "grad_norm": 0.07488419115543365, + "learning_rate": 2.739923396589693e-05, + "loss": 0.0209, + "step": 2770 + }, + { + "epoch": 4.656616415410385, + "grad_norm": 0.11510682106018066, + "learning_rate": 2.737592395871713e-05, + "loss": 0.0184, + "step": 2780 + }, + { + "epoch": 4.673366834170855, + "grad_norm": 0.08309126645326614, + "learning_rate": 2.7352519964621146e-05, + "loss": 0.0175, + "step": 2790 + }, + { + "epoch": 4.690117252931323, + "grad_norm": 0.0989244133234024, + "learning_rate": 2.7329022161346758e-05, + "loss": 0.02, + "step": 2800 + }, + { + "epoch": 4.706867671691793, + "grad_norm": 0.09160429984331131, + "learning_rate": 2.730543072734417e-05, + "loss": 0.0211, + "step": 2810 + }, + { + "epoch": 4.723618090452261, + "grad_norm": 0.13737095892429352, + "learning_rate": 2.728174584177463e-05, + "loss": 0.02, + "step": 2820 + }, + { + "epoch": 4.740368509212731, + "grad_norm": 0.09824742376804352, + "learning_rate": 2.725796768450911e-05, + "loss": 0.0187, + "step": 2830 + }, + { + "epoch": 4.757118927973199, + "grad_norm": 0.07674989849328995, + "learning_rate": 2.7234096436126903e-05, + "loss": 0.0201, + "step": 2840 + }, + { + "epoch": 4.773869346733669, + "grad_norm": 0.09987271577119827, + "learning_rate": 2.721013227791428e-05, + "loss": 0.0201, + "step": 2850 + }, + { + "epoch": 4.790619765494137, + "grad_norm": 0.10610336065292358, + "learning_rate": 2.7186075391863088e-05, + "loss": 0.0216, + "step": 2860 + }, + { + "epoch": 4.807370184254607, + "grad_norm": 0.09385973960161209, + "learning_rate": 2.7161925960669383e-05, + "loss": 0.0218, + "step": 2870 + }, + { + "epoch": 4.824120603015075, + "grad_norm": 0.1261572688817978, + "learning_rate": 2.713768416773205e-05, + "loss": 0.0201, + "step": 2880 + }, + { + "epoch": 4.840871021775545, + "grad_norm": 0.06238450109958649, + "learning_rate": 2.7113350197151394e-05, + "loss": 0.0192, + "step": 2890 + }, + { + "epoch": 4.857621440536013, + "grad_norm": 0.08462442457675934, + "learning_rate": 2.708892423372774e-05, + "loss": 0.0189, + "step": 2900 + }, + { + "epoch": 4.874371859296483, + "grad_norm": 0.11602508276700974, + "learning_rate": 2.7064406462960044e-05, + "loss": 0.0175, + "step": 2910 + }, + { + "epoch": 4.891122278056951, + "grad_norm": 0.08591975271701813, + "learning_rate": 2.7039797071044482e-05, + "loss": 0.023, + "step": 2920 + }, + { + "epoch": 4.907872696817421, + "grad_norm": 0.11215902119874954, + "learning_rate": 2.7015096244873028e-05, + "loss": 0.0204, + "step": 2930 + }, + { + "epoch": 4.924623115577889, + "grad_norm": 0.11343713849782944, + "learning_rate": 2.6990304172032038e-05, + "loss": 0.0199, + "step": 2940 + }, + { + "epoch": 4.941373534338359, + "grad_norm": 0.08644749969244003, + "learning_rate": 2.6965421040800823e-05, + "loss": 0.0219, + "step": 2950 + }, + { + "epoch": 4.958123953098827, + "grad_norm": 0.0927010178565979, + "learning_rate": 2.6940447040150228e-05, + "loss": 0.0194, + "step": 2960 + }, + { + "epoch": 4.974874371859297, + "grad_norm": 0.10158302634954453, + "learning_rate": 2.691538235974118e-05, + "loss": 0.0205, + "step": 2970 + }, + { + "epoch": 4.991624790619765, + "grad_norm": 0.10737495869398117, + "learning_rate": 2.6890227189923285e-05, + "loss": 0.0189, + "step": 2980 + }, + { + "epoch": 5.008375209380235, + "grad_norm": 0.07879546284675598, + "learning_rate": 2.6864981721733317e-05, + "loss": 0.0179, + "step": 2990 + }, + { + "epoch": 5.025125628140704, + "grad_norm": 0.10061459243297577, + "learning_rate": 2.683964614689383e-05, + "loss": 0.0196, + "step": 3000 + }, + { + "epoch": 5.041876046901173, + "grad_norm": 0.05866040289402008, + "learning_rate": 2.681422065781169e-05, + "loss": 0.0156, + "step": 3010 + }, + { + "epoch": 5.058626465661642, + "grad_norm": 0.13595989346504211, + "learning_rate": 2.6788705447576563e-05, + "loss": 0.0185, + "step": 3020 + }, + { + "epoch": 5.075376884422111, + "grad_norm": 0.10418418049812317, + "learning_rate": 2.6763100709959524e-05, + "loss": 0.0172, + "step": 3030 + }, + { + "epoch": 5.09212730318258, + "grad_norm": 0.09122584760189056, + "learning_rate": 2.6737406639411537e-05, + "loss": 0.0193, + "step": 3040 + }, + { + "epoch": 5.108877721943049, + "grad_norm": 0.10324976593255997, + "learning_rate": 2.6711623431061985e-05, + "loss": 0.0171, + "step": 3050 + }, + { + "epoch": 5.125628140703517, + "grad_norm": 0.11300911009311676, + "learning_rate": 2.6685751280717195e-05, + "loss": 0.0176, + "step": 3060 + }, + { + "epoch": 5.142378559463987, + "grad_norm": 0.08261276036500931, + "learning_rate": 2.665979038485896e-05, + "loss": 0.0195, + "step": 3070 + }, + { + "epoch": 5.159128978224456, + "grad_norm": 0.10486511141061783, + "learning_rate": 2.6633740940643022e-05, + "loss": 0.017, + "step": 3080 + }, + { + "epoch": 5.175879396984925, + "grad_norm": 0.10391178727149963, + "learning_rate": 2.6607603145897594e-05, + "loss": 0.0201, + "step": 3090 + }, + { + "epoch": 5.192629815745394, + "grad_norm": 0.0923885777592659, + "learning_rate": 2.6581377199121853e-05, + "loss": 0.0191, + "step": 3100 + }, + { + "epoch": 5.209380234505863, + "grad_norm": 0.12680231034755707, + "learning_rate": 2.6555063299484434e-05, + "loss": 0.0177, + "step": 3110 + }, + { + "epoch": 5.226130653266332, + "grad_norm": 0.07475026696920395, + "learning_rate": 2.6528661646821917e-05, + "loss": 0.0193, + "step": 3120 + }, + { + "epoch": 5.242881072026801, + "grad_norm": 0.10191482305526733, + "learning_rate": 2.6502172441637302e-05, + "loss": 0.0193, + "step": 3130 + }, + { + "epoch": 5.259631490787269, + "grad_norm": 0.08599104732275009, + "learning_rate": 2.6475595885098493e-05, + "loss": 0.0193, + "step": 3140 + }, + { + "epoch": 5.276381909547739, + "grad_norm": 0.0979737862944603, + "learning_rate": 2.644893217903677e-05, + "loss": 0.0177, + "step": 3150 + }, + { + "epoch": 5.293132328308207, + "grad_norm": 0.1444755345582962, + "learning_rate": 2.6422181525945265e-05, + "loss": 0.0185, + "step": 3160 + }, + { + "epoch": 5.309882747068677, + "grad_norm": 0.0918390303850174, + "learning_rate": 2.63953441289774e-05, + "loss": 0.02, + "step": 3170 + }, + { + "epoch": 5.326633165829146, + "grad_norm": 0.11050040274858475, + "learning_rate": 2.6368420191945365e-05, + "loss": 0.0181, + "step": 3180 + }, + { + "epoch": 5.343383584589615, + "grad_norm": 0.0900280773639679, + "learning_rate": 2.6341409919318572e-05, + "loss": 0.017, + "step": 3190 + }, + { + "epoch": 5.360134003350084, + "grad_norm": 0.08904273808002472, + "learning_rate": 2.6314313516222084e-05, + "loss": 0.0195, + "step": 3200 + }, + { + "epoch": 5.376884422110553, + "grad_norm": 0.10394366830587387, + "learning_rate": 2.6287131188435075e-05, + "loss": 0.0193, + "step": 3210 + }, + { + "epoch": 5.393634840871022, + "grad_norm": 0.10032975673675537, + "learning_rate": 2.6259863142389252e-05, + "loss": 0.0164, + "step": 3220 + }, + { + "epoch": 5.410385259631491, + "grad_norm": 0.09366985410451889, + "learning_rate": 2.62325095851673e-05, + "loss": 0.0171, + "step": 3230 + }, + { + "epoch": 5.42713567839196, + "grad_norm": 0.08237487822771072, + "learning_rate": 2.620507072450131e-05, + "loss": 0.0191, + "step": 3240 + }, + { + "epoch": 5.443886097152429, + "grad_norm": 0.10653271526098251, + "learning_rate": 2.6177546768771183e-05, + "loss": 0.0173, + "step": 3250 + }, + { + "epoch": 5.460636515912898, + "grad_norm": 0.08800845593214035, + "learning_rate": 2.6149937927003062e-05, + "loss": 0.018, + "step": 3260 + }, + { + "epoch": 5.477386934673367, + "grad_norm": 0.10618974268436432, + "learning_rate": 2.6122244408867766e-05, + "loss": 0.0198, + "step": 3270 + }, + { + "epoch": 5.494137353433836, + "grad_norm": 0.0838366374373436, + "learning_rate": 2.6094466424679148e-05, + "loss": 0.0196, + "step": 3280 + }, + { + "epoch": 5.510887772194305, + "grad_norm": 0.08228679746389389, + "learning_rate": 2.606660418539253e-05, + "loss": 0.0208, + "step": 3290 + }, + { + "epoch": 5.527638190954773, + "grad_norm": 0.10232897847890854, + "learning_rate": 2.6038657902603107e-05, + "loss": 0.0173, + "step": 3300 + }, + { + "epoch": 5.544388609715243, + "grad_norm": 0.0675511434674263, + "learning_rate": 2.6010627788544314e-05, + "loss": 0.0171, + "step": 3310 + }, + { + "epoch": 5.561139028475711, + "grad_norm": 0.11169396340847015, + "learning_rate": 2.5982514056086238e-05, + "loss": 0.0202, + "step": 3320 + }, + { + "epoch": 5.577889447236181, + "grad_norm": 0.09199520200490952, + "learning_rate": 2.5954316918733992e-05, + "loss": 0.0188, + "step": 3330 + }, + { + "epoch": 5.594639865996649, + "grad_norm": 0.09636520594358444, + "learning_rate": 2.5926036590626093e-05, + "loss": 0.0189, + "step": 3340 + }, + { + "epoch": 5.611390284757119, + "grad_norm": 0.1068258136510849, + "learning_rate": 2.589767328653283e-05, + "loss": 0.0185, + "step": 3350 + }, + { + "epoch": 5.628140703517588, + "grad_norm": 0.0738004669547081, + "learning_rate": 2.5869227221854646e-05, + "loss": 0.0153, + "step": 3360 + }, + { + "epoch": 5.644891122278057, + "grad_norm": 0.08421333134174347, + "learning_rate": 2.584069861262049e-05, + "loss": 0.0199, + "step": 3370 + }, + { + "epoch": 5.661641541038526, + "grad_norm": 0.1111551895737648, + "learning_rate": 2.58120876754862e-05, + "loss": 0.0147, + "step": 3380 + }, + { + "epoch": 5.678391959798995, + "grad_norm": 0.0835203304886818, + "learning_rate": 2.5783394627732815e-05, + "loss": 0.0168, + "step": 3390 + }, + { + "epoch": 5.695142378559464, + "grad_norm": 0.12761108577251434, + "learning_rate": 2.5754619687264957e-05, + "loss": 0.0189, + "step": 3400 + }, + { + "epoch": 5.711892797319933, + "grad_norm": 0.0886048823595047, + "learning_rate": 2.5725763072609178e-05, + "loss": 0.0164, + "step": 3410 + }, + { + "epoch": 5.728643216080402, + "grad_norm": 0.09086675196886063, + "learning_rate": 2.569682500291229e-05, + "loss": 0.0151, + "step": 3420 + }, + { + "epoch": 5.745393634840871, + "grad_norm": 0.10654979199171066, + "learning_rate": 2.5667805697939687e-05, + "loss": 0.0173, + "step": 3430 + }, + { + "epoch": 5.76214405360134, + "grad_norm": 0.09250108897686005, + "learning_rate": 2.5638705378073704e-05, + "loss": 0.0207, + "step": 3440 + }, + { + "epoch": 5.778894472361809, + "grad_norm": 0.09105037152767181, + "learning_rate": 2.5609524264311934e-05, + "loss": 0.0175, + "step": 3450 + }, + { + "epoch": 5.795644891122278, + "grad_norm": 0.10493557155132294, + "learning_rate": 2.5580262578265536e-05, + "loss": 0.0217, + "step": 3460 + }, + { + "epoch": 5.812395309882747, + "grad_norm": 0.09852892905473709, + "learning_rate": 2.5550920542157557e-05, + "loss": 0.0153, + "step": 3470 + }, + { + "epoch": 5.8291457286432165, + "grad_norm": 0.09802744537591934, + "learning_rate": 2.5521498378821273e-05, + "loss": 0.0195, + "step": 3480 + }, + { + "epoch": 5.8458961474036855, + "grad_norm": 0.07921155542135239, + "learning_rate": 2.549199631169844e-05, + "loss": 0.0171, + "step": 3490 + }, + { + "epoch": 5.8626465661641545, + "grad_norm": 0.09621422737836838, + "learning_rate": 2.5462414564837663e-05, + "loss": 0.0168, + "step": 3500 + }, + { + "epoch": 5.8793969849246235, + "grad_norm": 0.0989849641919136, + "learning_rate": 2.543275336289264e-05, + "loss": 0.017, + "step": 3510 + }, + { + "epoch": 5.8961474036850925, + "grad_norm": 0.11342565715312958, + "learning_rate": 2.540301293112048e-05, + "loss": 0.018, + "step": 3520 + }, + { + "epoch": 5.9128978224455615, + "grad_norm": 0.1399437040090561, + "learning_rate": 2.537319349538e-05, + "loss": 0.0182, + "step": 3530 + }, + { + "epoch": 5.9296482412060305, + "grad_norm": 0.09442800283432007, + "learning_rate": 2.534329528212999e-05, + "loss": 0.0164, + "step": 3540 + }, + { + "epoch": 5.9463986599664995, + "grad_norm": 0.09357256442308426, + "learning_rate": 2.531331851842751e-05, + "loss": 0.017, + "step": 3550 + }, + { + "epoch": 5.9631490787269685, + "grad_norm": 0.10102587193250656, + "learning_rate": 2.528326343192615e-05, + "loss": 0.0161, + "step": 3560 + }, + { + "epoch": 5.9798994974874375, + "grad_norm": 0.09273260831832886, + "learning_rate": 2.5253130250874316e-05, + "loss": 0.0168, + "step": 3570 + }, + { + "epoch": 5.9966499162479066, + "grad_norm": 0.09276672452688217, + "learning_rate": 2.5222919204113494e-05, + "loss": 0.0198, + "step": 3580 + }, + { + "epoch": 6.013400335008376, + "grad_norm": 0.15660251677036285, + "learning_rate": 2.5192630521076494e-05, + "loss": 0.0171, + "step": 3590 + }, + { + "epoch": 6.030150753768845, + "grad_norm": 0.10578886419534683, + "learning_rate": 2.5162264431785734e-05, + "loss": 0.0199, + "step": 3600 + }, + { + "epoch": 6.046901172529314, + "grad_norm": 0.10759575664997101, + "learning_rate": 2.5131821166851484e-05, + "loss": 0.0167, + "step": 3610 + }, + { + "epoch": 6.063651591289783, + "grad_norm": 0.08797171711921692, + "learning_rate": 2.5101300957470095e-05, + "loss": 0.0177, + "step": 3620 + }, + { + "epoch": 6.080402010050252, + "grad_norm": 0.06793040037155151, + "learning_rate": 2.507070403542228e-05, + "loss": 0.0166, + "step": 3630 + }, + { + "epoch": 6.097152428810721, + "grad_norm": 0.08354511857032776, + "learning_rate": 2.504003063307132e-05, + "loss": 0.0177, + "step": 3640 + }, + { + "epoch": 6.11390284757119, + "grad_norm": 0.06704747676849365, + "learning_rate": 2.500928098336131e-05, + "loss": 0.0184, + "step": 3650 + }, + { + "epoch": 6.130653266331659, + "grad_norm": 0.122153140604496, + "learning_rate": 2.497845531981541e-05, + "loss": 0.0191, + "step": 3660 + }, + { + "epoch": 6.147403685092128, + "grad_norm": 0.08629975467920303, + "learning_rate": 2.4947553876534047e-05, + "loss": 0.0175, + "step": 3670 + }, + { + "epoch": 6.164154103852597, + "grad_norm": 0.09296640753746033, + "learning_rate": 2.4916576888193137e-05, + "loss": 0.0148, + "step": 3680 + }, + { + "epoch": 6.180904522613066, + "grad_norm": 0.06364050507545471, + "learning_rate": 2.488552459004231e-05, + "loss": 0.0169, + "step": 3690 + }, + { + "epoch": 6.197654941373535, + "grad_norm": 0.07617541402578354, + "learning_rate": 2.4854397217903145e-05, + "loss": 0.0176, + "step": 3700 + }, + { + "epoch": 6.214405360134004, + "grad_norm": 0.08891832083463669, + "learning_rate": 2.4823195008167325e-05, + "loss": 0.0172, + "step": 3710 + }, + { + "epoch": 6.231155778894473, + "grad_norm": 0.06869089603424072, + "learning_rate": 2.4791918197794907e-05, + "loss": 0.0161, + "step": 3720 + }, + { + "epoch": 6.247906197654942, + "grad_norm": 0.10788177698850632, + "learning_rate": 2.4760567024312476e-05, + "loss": 0.0154, + "step": 3730 + }, + { + "epoch": 6.264656616415411, + "grad_norm": 0.14439617097377777, + "learning_rate": 2.4729141725811344e-05, + "loss": 0.0183, + "step": 3740 + }, + { + "epoch": 6.28140703517588, + "grad_norm": 0.10388529300689697, + "learning_rate": 2.4697642540945774e-05, + "loss": 0.0194, + "step": 3750 + }, + { + "epoch": 6.298157453936349, + "grad_norm": 0.08002973347902298, + "learning_rate": 2.4666069708931136e-05, + "loss": 0.015, + "step": 3760 + }, + { + "epoch": 6.314907872696818, + "grad_norm": 0.0883835032582283, + "learning_rate": 2.46344234695421e-05, + "loss": 0.0183, + "step": 3770 + }, + { + "epoch": 6.331658291457287, + "grad_norm": 0.08702576160430908, + "learning_rate": 2.460270406311082e-05, + "loss": 0.0159, + "step": 3780 + }, + { + "epoch": 6.348408710217756, + "grad_norm": 0.09502051025629044, + "learning_rate": 2.457091173052511e-05, + "loss": 0.0173, + "step": 3790 + }, + { + "epoch": 6.365159128978225, + "grad_norm": 0.09019004553556442, + "learning_rate": 2.4539046713226593e-05, + "loss": 0.0192, + "step": 3800 + }, + { + "epoch": 6.381909547738694, + "grad_norm": 0.10312219709157944, + "learning_rate": 2.4507109253208898e-05, + "loss": 0.0154, + "step": 3810 + }, + { + "epoch": 6.398659966499163, + "grad_norm": 0.08851608633995056, + "learning_rate": 2.4475099593015813e-05, + "loss": 0.0172, + "step": 3820 + }, + { + "epoch": 6.415410385259632, + "grad_norm": 0.0891028419137001, + "learning_rate": 2.4443017975739423e-05, + "loss": 0.0188, + "step": 3830 + }, + { + "epoch": 6.432160804020101, + "grad_norm": 0.08572867512702942, + "learning_rate": 2.4410864645018288e-05, + "loss": 0.0165, + "step": 3840 + }, + { + "epoch": 6.44891122278057, + "grad_norm": 0.09533847123384476, + "learning_rate": 2.4378639845035583e-05, + "loss": 0.0158, + "step": 3850 + }, + { + "epoch": 6.465661641541039, + "grad_norm": 0.0869334265589714, + "learning_rate": 2.4346343820517237e-05, + "loss": 0.0158, + "step": 3860 + }, + { + "epoch": 6.482412060301508, + "grad_norm": 0.10855986922979355, + "learning_rate": 2.4313976816730095e-05, + "loss": 0.0179, + "step": 3870 + }, + { + "epoch": 6.499162479061977, + "grad_norm": 0.0890968069434166, + "learning_rate": 2.4281539079480047e-05, + "loss": 0.0166, + "step": 3880 + }, + { + "epoch": 6.515912897822446, + "grad_norm": 0.09087546169757843, + "learning_rate": 2.4249030855110128e-05, + "loss": 0.0156, + "step": 3890 + }, + { + "epoch": 6.532663316582915, + "grad_norm": 0.07741452008485794, + "learning_rate": 2.42164523904987e-05, + "loss": 0.0163, + "step": 3900 + }, + { + "epoch": 6.549413735343384, + "grad_norm": 0.09288477897644043, + "learning_rate": 2.4183803933057554e-05, + "loss": 0.0165, + "step": 3910 + }, + { + "epoch": 6.566164154103853, + "grad_norm": 0.08580487966537476, + "learning_rate": 2.4151085730730013e-05, + "loss": 0.0201, + "step": 3920 + }, + { + "epoch": 6.582914572864322, + "grad_norm": 0.079682357609272, + "learning_rate": 2.4118298031989084e-05, + "loss": 0.0163, + "step": 3930 + }, + { + "epoch": 6.599664991624791, + "grad_norm": 0.07965271174907684, + "learning_rate": 2.4085441085835534e-05, + "loss": 0.0162, + "step": 3940 + }, + { + "epoch": 6.61641541038526, + "grad_norm": 0.10382688790559769, + "learning_rate": 2.405251514179604e-05, + "loss": 0.0172, + "step": 3950 + }, + { + "epoch": 6.633165829145729, + "grad_norm": 0.0887349396944046, + "learning_rate": 2.401952044992126e-05, + "loss": 0.0181, + "step": 3960 + }, + { + "epoch": 6.649916247906198, + "grad_norm": 0.10867097228765488, + "learning_rate": 2.3986457260783942e-05, + "loss": 0.016, + "step": 3970 + }, + { + "epoch": 6.666666666666667, + "grad_norm": 0.10828111320734024, + "learning_rate": 2.395332582547703e-05, + "loss": 0.0216, + "step": 3980 + }, + { + "epoch": 6.683417085427136, + "grad_norm": 0.08616909384727478, + "learning_rate": 2.3920126395611755e-05, + "loss": 0.019, + "step": 3990 + }, + { + "epoch": 6.700167504187605, + "grad_norm": 0.10397583991289139, + "learning_rate": 2.3886859223315725e-05, + "loss": 0.0199, + "step": 4000 + }, + { + "epoch": 6.716917922948074, + "grad_norm": 0.12109193205833435, + "learning_rate": 2.385352456123099e-05, + "loss": 0.0191, + "step": 4010 + }, + { + "epoch": 6.733668341708543, + "grad_norm": 0.07098208367824554, + "learning_rate": 2.3820122662512155e-05, + "loss": 0.0164, + "step": 4020 + }, + { + "epoch": 6.750418760469012, + "grad_norm": 0.1095917746424675, + "learning_rate": 2.3786653780824443e-05, + "loss": 0.0153, + "step": 4030 + }, + { + "epoch": 6.767169179229481, + "grad_norm": 0.08910483866930008, + "learning_rate": 2.3753118170341762e-05, + "loss": 0.0168, + "step": 4040 + }, + { + "epoch": 6.78391959798995, + "grad_norm": 0.07048118114471436, + "learning_rate": 2.3719516085744782e-05, + "loss": 0.0152, + "step": 4050 + }, + { + "epoch": 6.800670016750419, + "grad_norm": 0.1071673035621643, + "learning_rate": 2.3685847782218997e-05, + "loss": 0.0158, + "step": 4060 + }, + { + "epoch": 6.817420435510888, + "grad_norm": 0.10280796140432358, + "learning_rate": 2.3652113515452792e-05, + "loss": 0.0173, + "step": 4070 + }, + { + "epoch": 6.834170854271357, + "grad_norm": 0.10277356952428818, + "learning_rate": 2.3618313541635506e-05, + "loss": 0.0173, + "step": 4080 + }, + { + "epoch": 6.850921273031826, + "grad_norm": 0.09373802691698074, + "learning_rate": 2.358444811745547e-05, + "loss": 0.0166, + "step": 4090 + }, + { + "epoch": 6.867671691792295, + "grad_norm": 0.09374375641345978, + "learning_rate": 2.3550517500098068e-05, + "loss": 0.0171, + "step": 4100 + }, + { + "epoch": 6.884422110552764, + "grad_norm": 0.08589532971382141, + "learning_rate": 2.3516521947243787e-05, + "loss": 0.0177, + "step": 4110 + }, + { + "epoch": 6.901172529313233, + "grad_norm": 0.09759582579135895, + "learning_rate": 2.348246171706626e-05, + "loss": 0.0161, + "step": 4120 + }, + { + "epoch": 6.917922948073702, + "grad_norm": 0.08552340418100357, + "learning_rate": 2.3448337068230288e-05, + "loss": 0.0156, + "step": 4130 + }, + { + "epoch": 6.934673366834171, + "grad_norm": 0.07144278287887573, + "learning_rate": 2.3414148259889896e-05, + "loss": 0.0184, + "step": 4140 + }, + { + "epoch": 6.95142378559464, + "grad_norm": 0.08075377345085144, + "learning_rate": 2.3379895551686358e-05, + "loss": 0.0182, + "step": 4150 + }, + { + "epoch": 6.968174204355109, + "grad_norm": 0.09423919767141342, + "learning_rate": 2.334557920374623e-05, + "loss": 0.0173, + "step": 4160 + }, + { + "epoch": 6.984924623115578, + "grad_norm": 0.09013880789279938, + "learning_rate": 2.3311199476679367e-05, + "loss": 0.0172, + "step": 4170 + }, + { + "epoch": 7.001675041876047, + "grad_norm": 0.1004585400223732, + "learning_rate": 2.327675663157693e-05, + "loss": 0.0171, + "step": 4180 + }, + { + "epoch": 7.018425460636516, + "grad_norm": 0.09630786627531052, + "learning_rate": 2.3242250930009446e-05, + "loss": 0.0169, + "step": 4190 + }, + { + "epoch": 7.035175879396985, + "grad_norm": 0.09441886842250824, + "learning_rate": 2.3207682634024777e-05, + "loss": 0.016, + "step": 4200 + }, + { + "epoch": 7.051926298157454, + "grad_norm": 0.1110304594039917, + "learning_rate": 2.3173052006146155e-05, + "loss": 0.0163, + "step": 4210 + }, + { + "epoch": 7.068676716917923, + "grad_norm": 0.07521197199821472, + "learning_rate": 2.313835930937018e-05, + "loss": 0.0146, + "step": 4220 + }, + { + "epoch": 7.085427135678392, + "grad_norm": 0.06828752905130386, + "learning_rate": 2.310360480716483e-05, + "loss": 0.0156, + "step": 4230 + }, + { + "epoch": 7.102177554438861, + "grad_norm": 0.08863577991724014, + "learning_rate": 2.3068788763467434e-05, + "loss": 0.0154, + "step": 4240 + }, + { + "epoch": 7.11892797319933, + "grad_norm": 0.10143563151359558, + "learning_rate": 2.3033911442682708e-05, + "loss": 0.0171, + "step": 4250 + }, + { + "epoch": 7.135678391959799, + "grad_norm": 0.10039176791906357, + "learning_rate": 2.299897310968072e-05, + "loss": 0.0185, + "step": 4260 + }, + { + "epoch": 7.152428810720268, + "grad_norm": 0.06830760836601257, + "learning_rate": 2.2963974029794893e-05, + "loss": 0.0135, + "step": 4270 + }, + { + "epoch": 7.169179229480737, + "grad_norm": 0.06696446985006332, + "learning_rate": 2.2928914468819968e-05, + "loss": 0.016, + "step": 4280 + }, + { + "epoch": 7.185929648241206, + "grad_norm": 0.06527981907129288, + "learning_rate": 2.2893794693010002e-05, + "loss": 0.0149, + "step": 4290 + }, + { + "epoch": 7.202680067001675, + "grad_norm": 0.08876840025186539, + "learning_rate": 2.2858614969076358e-05, + "loss": 0.0179, + "step": 4300 + }, + { + "epoch": 7.219430485762144, + "grad_norm": 0.08040497452020645, + "learning_rate": 2.2823375564185645e-05, + "loss": 0.0172, + "step": 4310 + }, + { + "epoch": 7.236180904522613, + "grad_norm": 0.06119128316640854, + "learning_rate": 2.278807674595773e-05, + "loss": 0.0157, + "step": 4320 + }, + { + "epoch": 7.252931323283082, + "grad_norm": 0.07552856206893921, + "learning_rate": 2.275271878246367e-05, + "loss": 0.0161, + "step": 4330 + }, + { + "epoch": 7.269681742043551, + "grad_norm": 0.09174931794404984, + "learning_rate": 2.271730194222369e-05, + "loss": 0.0157, + "step": 4340 + }, + { + "epoch": 7.28643216080402, + "grad_norm": 0.10719872266054153, + "learning_rate": 2.2681826494205156e-05, + "loss": 0.0175, + "step": 4350 + }, + { + "epoch": 7.303182579564489, + "grad_norm": 0.08944938331842422, + "learning_rate": 2.2646292707820513e-05, + "loss": 0.0146, + "step": 4360 + }, + { + "epoch": 7.319932998324958, + "grad_norm": 0.08042101562023163, + "learning_rate": 2.2610700852925254e-05, + "loss": 0.0145, + "step": 4370 + }, + { + "epoch": 7.336683417085427, + "grad_norm": 0.10604419559240341, + "learning_rate": 2.2575051199815854e-05, + "loss": 0.0175, + "step": 4380 + }, + { + "epoch": 7.353433835845896, + "grad_norm": 0.07255196571350098, + "learning_rate": 2.253934401922773e-05, + "loss": 0.0161, + "step": 4390 + }, + { + "epoch": 7.370184254606365, + "grad_norm": 0.07271875441074371, + "learning_rate": 2.250357958233319e-05, + "loss": 0.0181, + "step": 4400 + }, + { + "epoch": 7.386934673366834, + "grad_norm": 0.08586633205413818, + "learning_rate": 2.2467758160739352e-05, + "loss": 0.0159, + "step": 4410 + }, + { + "epoch": 7.403685092127303, + "grad_norm": 0.07192263007164001, + "learning_rate": 2.2431880026486114e-05, + "loss": 0.0176, + "step": 4420 + }, + { + "epoch": 7.420435510887772, + "grad_norm": 0.0814385861158371, + "learning_rate": 2.2395945452044045e-05, + "loss": 0.0171, + "step": 4430 + }, + { + "epoch": 7.437185929648241, + "grad_norm": 0.0670303925871849, + "learning_rate": 2.2359954710312363e-05, + "loss": 0.0145, + "step": 4440 + }, + { + "epoch": 7.45393634840871, + "grad_norm": 0.08451966196298599, + "learning_rate": 2.232390807461682e-05, + "loss": 0.0148, + "step": 4450 + }, + { + "epoch": 7.4706867671691795, + "grad_norm": 0.09510994702577591, + "learning_rate": 2.2287805818707658e-05, + "loss": 0.016, + "step": 4460 + }, + { + "epoch": 7.4874371859296485, + "grad_norm": 0.08456354588270187, + "learning_rate": 2.2251648216757514e-05, + "loss": 0.014, + "step": 4470 + }, + { + "epoch": 7.5041876046901175, + "grad_norm": 0.11731875687837601, + "learning_rate": 2.2215435543359337e-05, + "loss": 0.0154, + "step": 4480 + }, + { + "epoch": 7.5209380234505865, + "grad_norm": 0.10068397969007492, + "learning_rate": 2.2179168073524308e-05, + "loss": 0.0161, + "step": 4490 + }, + { + "epoch": 7.5376884422110555, + "grad_norm": 0.09683054685592651, + "learning_rate": 2.2142846082679763e-05, + "loss": 0.0168, + "step": 4500 + }, + { + "epoch": 7.5544388609715245, + "grad_norm": 0.06540673226118088, + "learning_rate": 2.2106469846667078e-05, + "loss": 0.013, + "step": 4510 + }, + { + "epoch": 7.5711892797319935, + "grad_norm": 0.0816856399178505, + "learning_rate": 2.2070039641739578e-05, + "loss": 0.0154, + "step": 4520 + }, + { + "epoch": 7.5879396984924625, + "grad_norm": 0.07817541062831879, + "learning_rate": 2.2033555744560462e-05, + "loss": 0.017, + "step": 4530 + }, + { + "epoch": 7.6046901172529315, + "grad_norm": 0.08995489031076431, + "learning_rate": 2.1997018432200674e-05, + "loss": 0.0153, + "step": 4540 + }, + { + "epoch": 7.6214405360134005, + "grad_norm": 0.09150712937116623, + "learning_rate": 2.1960427982136818e-05, + "loss": 0.0154, + "step": 4550 + }, + { + "epoch": 7.63819095477387, + "grad_norm": 0.11141885071992874, + "learning_rate": 2.1923784672249038e-05, + "loss": 0.0159, + "step": 4560 + }, + { + "epoch": 7.654941373534339, + "grad_norm": 0.09261714667081833, + "learning_rate": 2.1887088780818918e-05, + "loss": 0.015, + "step": 4570 + }, + { + "epoch": 7.671691792294808, + "grad_norm": 0.11019844561815262, + "learning_rate": 2.1850340586527354e-05, + "loss": 0.0181, + "step": 4580 + }, + { + "epoch": 7.688442211055277, + "grad_norm": 0.09286614507436752, + "learning_rate": 2.1813540368452468e-05, + "loss": 0.0165, + "step": 4590 + }, + { + "epoch": 7.705192629815746, + "grad_norm": 0.07964478433132172, + "learning_rate": 2.1776688406067447e-05, + "loss": 0.0162, + "step": 4600 + }, + { + "epoch": 7.721943048576215, + "grad_norm": 0.09548459202051163, + "learning_rate": 2.1739784979238447e-05, + "loss": 0.0165, + "step": 4610 + }, + { + "epoch": 7.738693467336684, + "grad_norm": 0.06440986692905426, + "learning_rate": 2.1702830368222478e-05, + "loss": 0.0141, + "step": 4620 + }, + { + "epoch": 7.755443886097153, + "grad_norm": 0.07117027044296265, + "learning_rate": 2.166582485366523e-05, + "loss": 0.0147, + "step": 4630 + }, + { + "epoch": 7.772194304857622, + "grad_norm": 0.08931209146976471, + "learning_rate": 2.1628768716599e-05, + "loss": 0.0147, + "step": 4640 + }, + { + "epoch": 7.788944723618091, + "grad_norm": 0.09026764333248138, + "learning_rate": 2.1591662238440515e-05, + "loss": 0.0159, + "step": 4650 + }, + { + "epoch": 7.80569514237856, + "grad_norm": 0.1498933583498001, + "learning_rate": 2.1554505700988813e-05, + "loss": 0.0155, + "step": 4660 + }, + { + "epoch": 7.822445561139029, + "grad_norm": 0.06725804507732391, + "learning_rate": 2.1517299386423094e-05, + "loss": 0.016, + "step": 4670 + }, + { + "epoch": 7.839195979899498, + "grad_norm": 0.08815880864858627, + "learning_rate": 2.148004357730059e-05, + "loss": 0.0151, + "step": 4680 + }, + { + "epoch": 7.855946398659967, + "grad_norm": 0.06878040730953217, + "learning_rate": 2.1442738556554404e-05, + "loss": 0.0172, + "step": 4690 + }, + { + "epoch": 7.872696817420436, + "grad_norm": 0.06804270297288895, + "learning_rate": 2.1405384607491375e-05, + "loss": 0.0156, + "step": 4700 + }, + { + "epoch": 7.889447236180905, + "grad_norm": 0.070171058177948, + "learning_rate": 2.1367982013789913e-05, + "loss": 0.0153, + "step": 4710 + }, + { + "epoch": 7.906197654941374, + "grad_norm": 0.07167346775531769, + "learning_rate": 2.133053105949786e-05, + "loss": 0.0156, + "step": 4720 + }, + { + "epoch": 7.922948073701843, + "grad_norm": 0.09849503636360168, + "learning_rate": 2.129303202903032e-05, + "loss": 0.0165, + "step": 4730 + }, + { + "epoch": 7.939698492462312, + "grad_norm": 0.09628047794103622, + "learning_rate": 2.1255485207167504e-05, + "loss": 0.016, + "step": 4740 + }, + { + "epoch": 7.956448911222781, + "grad_norm": 0.07518996298313141, + "learning_rate": 2.1217890879052556e-05, + "loss": 0.0148, + "step": 4750 + }, + { + "epoch": 7.97319932998325, + "grad_norm": 0.09254742413759232, + "learning_rate": 2.1180249330189423e-05, + "loss": 0.0177, + "step": 4760 + }, + { + "epoch": 7.989949748743719, + "grad_norm": 0.10075143724679947, + "learning_rate": 2.1142560846440645e-05, + "loss": 0.0173, + "step": 4770 + }, + { + "epoch": 8.006700167504187, + "grad_norm": 0.0872223973274231, + "learning_rate": 2.1104825714025203e-05, + "loss": 0.0155, + "step": 4780 + }, + { + "epoch": 8.023450586264657, + "grad_norm": 0.07023284584283829, + "learning_rate": 2.106704421951634e-05, + "loss": 0.0169, + "step": 4790 + }, + { + "epoch": 8.040201005025125, + "grad_norm": 0.0736195370554924, + "learning_rate": 2.1029216649839404e-05, + "loss": 0.0153, + "step": 4800 + }, + { + "epoch": 8.056951423785595, + "grad_norm": 0.12984727323055267, + "learning_rate": 2.099134329226963e-05, + "loss": 0.0194, + "step": 4810 + }, + { + "epoch": 8.073701842546063, + "grad_norm": 0.06633501499891281, + "learning_rate": 2.0953424434430006e-05, + "loss": 0.0172, + "step": 4820 + }, + { + "epoch": 8.090452261306533, + "grad_norm": 0.07725346088409424, + "learning_rate": 2.091546036428905e-05, + "loss": 0.0167, + "step": 4830 + }, + { + "epoch": 8.107202680067001, + "grad_norm": 0.0776301696896553, + "learning_rate": 2.0877451370158633e-05, + "loss": 0.0167, + "step": 4840 + }, + { + "epoch": 8.12395309882747, + "grad_norm": 0.08715241402387619, + "learning_rate": 2.083939774069181e-05, + "loss": 0.0152, + "step": 4850 + }, + { + "epoch": 8.140703517587939, + "grad_norm": 0.05368698760867119, + "learning_rate": 2.080129976488059e-05, + "loss": 0.0158, + "step": 4860 + }, + { + "epoch": 8.157453936348409, + "grad_norm": 0.07227354496717453, + "learning_rate": 2.0763157732053795e-05, + "loss": 0.0167, + "step": 4870 + }, + { + "epoch": 8.174204355108877, + "grad_norm": 0.08462859690189362, + "learning_rate": 2.0724971931874796e-05, + "loss": 0.0155, + "step": 4880 + }, + { + "epoch": 8.190954773869347, + "grad_norm": 0.0951157882809639, + "learning_rate": 2.0686742654339363e-05, + "loss": 0.0151, + "step": 4890 + }, + { + "epoch": 8.207705192629815, + "grad_norm": 0.07463134825229645, + "learning_rate": 2.0648470189773455e-05, + "loss": 0.0136, + "step": 4900 + }, + { + "epoch": 8.224455611390285, + "grad_norm": 0.08849921822547913, + "learning_rate": 2.0610154828830997e-05, + "loss": 0.0145, + "step": 4910 + }, + { + "epoch": 8.241206030150753, + "grad_norm": 0.08144073933362961, + "learning_rate": 2.0571796862491686e-05, + "loss": 0.0173, + "step": 4920 + }, + { + "epoch": 8.257956448911223, + "grad_norm": 0.08935582637786865, + "learning_rate": 2.0533396582058783e-05, + "loss": 0.0173, + "step": 4930 + }, + { + "epoch": 8.274706867671691, + "grad_norm": 0.06497425585985184, + "learning_rate": 2.0494954279156893e-05, + "loss": 0.0159, + "step": 4940 + }, + { + "epoch": 8.291457286432161, + "grad_norm": 0.09040762484073639, + "learning_rate": 2.0456470245729748e-05, + "loss": 0.018, + "step": 4950 + }, + { + "epoch": 8.308207705192629, + "grad_norm": 0.11972896009683609, + "learning_rate": 2.0417944774038013e-05, + "loss": 0.016, + "step": 4960 + }, + { + "epoch": 8.324958123953099, + "grad_norm": 0.09560854732990265, + "learning_rate": 2.0379378156657026e-05, + "loss": 0.0151, + "step": 4970 + }, + { + "epoch": 8.341708542713567, + "grad_norm": 0.09106440097093582, + "learning_rate": 2.0340770686474618e-05, + "loss": 0.0169, + "step": 4980 + }, + { + "epoch": 8.358458961474037, + "grad_norm": 0.07660805433988571, + "learning_rate": 2.030212265668886e-05, + "loss": 0.0141, + "step": 4990 + }, + { + "epoch": 8.375209380234505, + "grad_norm": 0.08468059450387955, + "learning_rate": 2.0263434360805847e-05, + "loss": 0.0167, + "step": 5000 + } + ], + "logging_steps": 10, + "max_steps": 12000, + "num_input_tokens_seen": 0, + "num_train_epochs": 21, + "save_steps": 12001, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-8000/config.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-8000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..47d7b77e4255263f3803a4b6f94ca6831e459b70 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-8000/config.json @@ -0,0 +1,64 @@ +{ + "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" + ], + "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", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-8000/trainer_state.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-8000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..42be4a6de6ce06f4b97a5ecff84617f6533db5a4 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-8000/trainer_state.json @@ -0,0 +1,5634 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 13.40033500837521, + "eval_steps": 500, + "global_step": 8000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.01675041876046901, + "grad_norm": 1.1124801635742188, + "learning_rate": 4.5e-07, + "loss": 0.2576, + "step": 10 + }, + { + "epoch": 0.03350083752093802, + "grad_norm": 1.3764758110046387, + "learning_rate": 9.500000000000001e-07, + "loss": 0.2541, + "step": 20 + }, + { + "epoch": 0.05025125628140704, + "grad_norm": 1.0254155397415161, + "learning_rate": 1.45e-06, + "loss": 0.2369, + "step": 30 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 0.6008508801460266, + "learning_rate": 1.95e-06, + "loss": 0.2272, + "step": 40 + }, + { + "epoch": 0.08375209380234507, + "grad_norm": 1.085799217224121, + "learning_rate": 2.45e-06, + "loss": 0.2211, + "step": 50 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 0.9870124459266663, + "learning_rate": 2.9499999999999997e-06, + "loss": 0.2511, + "step": 60 + }, + { + "epoch": 0.11725293132328309, + "grad_norm": 0.9730237722396851, + "learning_rate": 3.4500000000000004e-06, + "loss": 0.2414, + "step": 70 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 0.792273998260498, + "learning_rate": 3.9499999999999995e-06, + "loss": 0.2311, + "step": 80 + }, + { + "epoch": 0.1507537688442211, + "grad_norm": 1.078325629234314, + "learning_rate": 4.450000000000001e-06, + "loss": 0.2185, + "step": 90 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 0.6237806081771851, + "learning_rate": 4.95e-06, + "loss": 0.1754, + "step": 100 + }, + { + "epoch": 0.18425460636515914, + "grad_norm": 0.7615986466407776, + "learning_rate": 5.45e-06, + "loss": 0.167, + "step": 110 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 0.6572969555854797, + "learning_rate": 5.95e-06, + "loss": 0.1568, + "step": 120 + }, + { + "epoch": 0.21775544388609716, + "grad_norm": 0.5472678542137146, + "learning_rate": 6.45e-06, + "loss": 0.1421, + "step": 130 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 0.4665081202983856, + "learning_rate": 6.95e-06, + "loss": 0.1526, + "step": 140 + }, + { + "epoch": 0.25125628140703515, + "grad_norm": 0.5054625272750854, + "learning_rate": 7.45e-06, + "loss": 0.1246, + "step": 150 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 0.5371837019920349, + "learning_rate": 7.95e-06, + "loss": 0.1179, + "step": 160 + }, + { + "epoch": 0.2847571189279732, + "grad_norm": 0.40661293268203735, + "learning_rate": 8.45e-06, + "loss": 0.1062, + "step": 170 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 0.2422139048576355, + "learning_rate": 8.95e-06, + "loss": 0.0963, + "step": 180 + }, + { + "epoch": 0.31825795644891125, + "grad_norm": 0.3034665584564209, + "learning_rate": 9.450000000000001e-06, + "loss": 0.0975, + "step": 190 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 0.2536943256855011, + "learning_rate": 9.95e-06, + "loss": 0.0934, + "step": 200 + }, + { + "epoch": 0.35175879396984927, + "grad_norm": 0.2668403685092926, + "learning_rate": 1.045e-05, + "loss": 0.0933, + "step": 210 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 0.22533497214317322, + "learning_rate": 1.095e-05, + "loss": 0.0792, + "step": 220 + }, + { + "epoch": 0.3852596314907873, + "grad_norm": 0.2040107697248459, + "learning_rate": 1.145e-05, + "loss": 0.0812, + "step": 230 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 0.19936227798461914, + "learning_rate": 1.1949999999999999e-05, + "loss": 0.0838, + "step": 240 + }, + { + "epoch": 0.4187604690117253, + "grad_norm": 0.2014681100845337, + "learning_rate": 1.245e-05, + "loss": 0.0804, + "step": 250 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 0.20712223649024963, + "learning_rate": 1.295e-05, + "loss": 0.0751, + "step": 260 + }, + { + "epoch": 0.45226130653266333, + "grad_norm": 0.23043803870677948, + "learning_rate": 1.345e-05, + "loss": 0.0747, + "step": 270 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 0.19819584488868713, + "learning_rate": 1.395e-05, + "loss": 0.0715, + "step": 280 + }, + { + "epoch": 0.48576214405360135, + "grad_norm": 0.2228541076183319, + "learning_rate": 1.445e-05, + "loss": 0.0708, + "step": 290 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 0.17452232539653778, + "learning_rate": 1.4950000000000001e-05, + "loss": 0.0685, + "step": 300 + }, + { + "epoch": 0.5192629815745393, + "grad_norm": 0.20126119256019592, + "learning_rate": 1.545e-05, + "loss": 0.0703, + "step": 310 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 0.2398366630077362, + "learning_rate": 1.5949999999999998e-05, + "loss": 0.0664, + "step": 320 + }, + { + "epoch": 0.5527638190954773, + "grad_norm": 0.21752075850963593, + "learning_rate": 1.645e-05, + "loss": 0.0679, + "step": 330 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 0.17013588547706604, + "learning_rate": 1.695e-05, + "loss": 0.0705, + "step": 340 + }, + { + "epoch": 0.5862646566164154, + "grad_norm": 0.2016950249671936, + "learning_rate": 1.745e-05, + "loss": 0.0687, + "step": 350 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 0.1707467883825302, + "learning_rate": 1.7950000000000003e-05, + "loss": 0.0675, + "step": 360 + }, + { + "epoch": 0.6197654941373534, + "grad_norm": 0.17180311679840088, + "learning_rate": 1.845e-05, + "loss": 0.0618, + "step": 370 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 0.21700109541416168, + "learning_rate": 1.8950000000000003e-05, + "loss": 0.0631, + "step": 380 + }, + { + "epoch": 0.6532663316582915, + "grad_norm": 0.21110156178474426, + "learning_rate": 1.945e-05, + "loss": 0.061, + "step": 390 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 0.17809593677520752, + "learning_rate": 1.995e-05, + "loss": 0.0585, + "step": 400 + }, + { + "epoch": 0.6867671691792295, + "grad_norm": 0.15697689354419708, + "learning_rate": 2.045e-05, + "loss": 0.0615, + "step": 410 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 0.17097905278205872, + "learning_rate": 2.095e-05, + "loss": 0.0568, + "step": 420 + }, + { + "epoch": 0.7202680067001676, + "grad_norm": 0.17003680765628815, + "learning_rate": 2.145e-05, + "loss": 0.0609, + "step": 430 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 0.20018595457077026, + "learning_rate": 2.195e-05, + "loss": 0.0538, + "step": 440 + }, + { + "epoch": 0.7537688442211056, + "grad_norm": 0.1872512400150299, + "learning_rate": 2.245e-05, + "loss": 0.0562, + "step": 450 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 0.210875004529953, + "learning_rate": 2.2950000000000002e-05, + "loss": 0.0538, + "step": 460 + }, + { + "epoch": 0.7872696817420436, + "grad_norm": 0.19456243515014648, + "learning_rate": 2.345e-05, + "loss": 0.0548, + "step": 470 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 0.19903069734573364, + "learning_rate": 2.395e-05, + "loss": 0.0536, + "step": 480 + }, + { + "epoch": 0.8207705192629816, + "grad_norm": 0.1998818963766098, + "learning_rate": 2.4449999999999998e-05, + "loss": 0.0533, + "step": 490 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 0.22372868657112122, + "learning_rate": 2.495e-05, + "loss": 0.0532, + "step": 500 + }, + { + "epoch": 0.8542713567839196, + "grad_norm": 0.2313748300075531, + "learning_rate": 2.5450000000000002e-05, + "loss": 0.049, + "step": 510 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 0.1595960259437561, + "learning_rate": 2.595e-05, + "loss": 0.0513, + "step": 520 + }, + { + "epoch": 0.8877721943048577, + "grad_norm": 0.3054298162460327, + "learning_rate": 2.6450000000000003e-05, + "loss": 0.0489, + "step": 530 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 0.15157485008239746, + "learning_rate": 2.695e-05, + "loss": 0.0501, + "step": 540 + }, + { + "epoch": 0.9212730318257957, + "grad_norm": 0.19291409850120544, + "learning_rate": 2.7450000000000003e-05, + "loss": 0.0428, + "step": 550 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 0.22616443037986755, + "learning_rate": 2.795e-05, + "loss": 0.048, + "step": 560 + }, + { + "epoch": 0.9547738693467337, + "grad_norm": 0.21530359983444214, + "learning_rate": 2.845e-05, + "loss": 0.051, + "step": 570 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 0.3027316629886627, + "learning_rate": 2.895e-05, + "loss": 0.0462, + "step": 580 + }, + { + "epoch": 0.9882747068676717, + "grad_norm": 0.16115690767765045, + "learning_rate": 2.945e-05, + "loss": 0.0438, + "step": 590 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 0.20157869160175323, + "learning_rate": 2.995e-05, + "loss": 0.0451, + "step": 600 + }, + { + "epoch": 1.0217755443886096, + "grad_norm": 0.19205620884895325, + "learning_rate": 2.9999953864407506e-05, + "loss": 0.0437, + "step": 610 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 0.1747637391090393, + "learning_rate": 2.99997943837114e-05, + "loss": 0.0454, + "step": 620 + }, + { + "epoch": 1.0552763819095476, + "grad_norm": 0.18132591247558594, + "learning_rate": 2.9999520989547326e-05, + "loss": 0.0488, + "step": 630 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 0.1378272920846939, + "learning_rate": 2.9999133683991525e-05, + "loss": 0.042, + "step": 640 + }, + { + "epoch": 1.0887772194304857, + "grad_norm": 0.17927886545658112, + "learning_rate": 2.9998632469985326e-05, + "loss": 0.0452, + "step": 650 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 0.1842069774866104, + "learning_rate": 2.999801735133512e-05, + "loss": 0.0418, + "step": 660 + }, + { + "epoch": 1.1222780569514237, + "grad_norm": 0.1894877851009369, + "learning_rate": 2.9997288332712325e-05, + "loss": 0.0421, + "step": 670 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 0.17502732574939728, + "learning_rate": 2.999644541965335e-05, + "loss": 0.0395, + "step": 680 + }, + { + "epoch": 1.1557788944723617, + "grad_norm": 0.19280853867530823, + "learning_rate": 2.9995488618559554e-05, + "loss": 0.0383, + "step": 690 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 0.21065080165863037, + "learning_rate": 2.9994417936697218e-05, + "loss": 0.0409, + "step": 700 + }, + { + "epoch": 1.1892797319932997, + "grad_norm": 0.15386411547660828, + "learning_rate": 2.9993233382197454e-05, + "loss": 0.0417, + "step": 710 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 0.18249723315238953, + "learning_rate": 2.999193496405616e-05, + "loss": 0.0386, + "step": 720 + }, + { + "epoch": 1.2227805695142377, + "grad_norm": 0.17197810113430023, + "learning_rate": 2.9990522692133962e-05, + "loss": 0.038, + "step": 730 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 0.15405157208442688, + "learning_rate": 2.9988996577156134e-05, + "loss": 0.0421, + "step": 740 + }, + { + "epoch": 1.2562814070351758, + "grad_norm": 0.16718165576457977, + "learning_rate": 2.9987356630712496e-05, + "loss": 0.04, + "step": 750 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 0.1574426144361496, + "learning_rate": 2.9985602865257354e-05, + "loss": 0.0403, + "step": 760 + }, + { + "epoch": 1.2897822445561138, + "grad_norm": 0.11435810476541519, + "learning_rate": 2.998373529410939e-05, + "loss": 0.0398, + "step": 770 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 0.16009581089019775, + "learning_rate": 2.9981753931451567e-05, + "loss": 0.037, + "step": 780 + }, + { + "epoch": 1.3232830820770518, + "grad_norm": 0.15292680263519287, + "learning_rate": 2.9979658792331016e-05, + "loss": 0.0375, + "step": 790 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 0.1592884510755539, + "learning_rate": 2.997744989265893e-05, + "loss": 0.037, + "step": 800 + }, + { + "epoch": 1.3567839195979898, + "grad_norm": 0.22414714097976685, + "learning_rate": 2.9975127249210426e-05, + "loss": 0.0344, + "step": 810 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 0.14030888676643372, + "learning_rate": 2.9972690879624445e-05, + "loss": 0.0364, + "step": 820 + }, + { + "epoch": 1.3902847571189278, + "grad_norm": 0.15426629781723022, + "learning_rate": 2.9970140802403587e-05, + "loss": 0.0368, + "step": 830 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 0.1377844214439392, + "learning_rate": 2.9967477036914002e-05, + "loss": 0.0365, + "step": 840 + }, + { + "epoch": 1.4237855946398659, + "grad_norm": 0.1314632147550583, + "learning_rate": 2.996469960338521e-05, + "loss": 0.0364, + "step": 850 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 0.1743793785572052, + "learning_rate": 2.9961808522909976e-05, + "loss": 0.0383, + "step": 860 + }, + { + "epoch": 1.457286432160804, + "grad_norm": 0.1816023737192154, + "learning_rate": 2.995880381744414e-05, + "loss": 0.0357, + "step": 870 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 0.19835101068019867, + "learning_rate": 2.995568550980643e-05, + "loss": 0.0383, + "step": 880 + }, + { + "epoch": 1.490787269681742, + "grad_norm": 0.13531947135925293, + "learning_rate": 2.9952453623678334e-05, + "loss": 0.0338, + "step": 890 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 0.13880357146263123, + "learning_rate": 2.9949108183603867e-05, + "loss": 0.036, + "step": 900 + }, + { + "epoch": 1.52428810720268, + "grad_norm": 0.18743862211704254, + "learning_rate": 2.994564921498943e-05, + "loss": 0.0358, + "step": 910 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 0.1355157345533371, + "learning_rate": 2.9942076744103594e-05, + "loss": 0.0342, + "step": 920 + }, + { + "epoch": 1.557788944723618, + "grad_norm": 0.16391365230083466, + "learning_rate": 2.9938390798076886e-05, + "loss": 0.0359, + "step": 930 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 0.1717372089624405, + "learning_rate": 2.993459140490163e-05, + "loss": 0.0344, + "step": 940 + }, + { + "epoch": 1.591289782244556, + "grad_norm": 0.19507038593292236, + "learning_rate": 2.9930678593431684e-05, + "loss": 0.0325, + "step": 950 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 0.14840291440486908, + "learning_rate": 2.9926652393382258e-05, + "loss": 0.0334, + "step": 960 + }, + { + "epoch": 1.624790619765494, + "grad_norm": 0.15482094883918762, + "learning_rate": 2.9922512835329655e-05, + "loss": 0.0338, + "step": 970 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 0.12408526986837387, + "learning_rate": 2.9918259950711073e-05, + "loss": 0.0324, + "step": 980 + }, + { + "epoch": 1.658291457286432, + "grad_norm": 0.1402246356010437, + "learning_rate": 2.991389377182435e-05, + "loss": 0.0363, + "step": 990 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 0.15740147233009338, + "learning_rate": 2.99094143318277e-05, + "loss": 0.0328, + "step": 1000 + }, + { + "epoch": 1.69179229480737, + "grad_norm": 0.14837925136089325, + "learning_rate": 2.9904821664739493e-05, + "loss": 0.0332, + "step": 1010 + }, + { + "epoch": 1.708542713567839, + "grad_norm": 0.1531188040971756, + "learning_rate": 2.990011580543799e-05, + "loss": 0.0326, + "step": 1020 + }, + { + "epoch": 1.725293132328308, + "grad_norm": 0.1466841697692871, + "learning_rate": 2.9895296789661065e-05, + "loss": 0.0326, + "step": 1030 + }, + { + "epoch": 1.742043551088777, + "grad_norm": 0.14380747079849243, + "learning_rate": 2.989036465400594e-05, + "loss": 0.0311, + "step": 1040 + }, + { + "epoch": 1.758793969849246, + "grad_norm": 0.15117715299129486, + "learning_rate": 2.9885319435928896e-05, + "loss": 0.0308, + "step": 1050 + }, + { + "epoch": 1.775544388609715, + "grad_norm": 0.1409190148115158, + "learning_rate": 2.9880161173745017e-05, + "loss": 0.0288, + "step": 1060 + }, + { + "epoch": 1.792294807370184, + "grad_norm": 0.11704489588737488, + "learning_rate": 2.9874889906627882e-05, + "loss": 0.0314, + "step": 1070 + }, + { + "epoch": 1.809045226130653, + "grad_norm": 0.1579095870256424, + "learning_rate": 2.9869505674609253e-05, + "loss": 0.0334, + "step": 1080 + }, + { + "epoch": 1.8257956448911221, + "grad_norm": 0.14541688561439514, + "learning_rate": 2.9864008518578794e-05, + "loss": 0.0301, + "step": 1090 + }, + { + "epoch": 1.8425460636515911, + "grad_norm": 0.11781132966279984, + "learning_rate": 2.9858398480283758e-05, + "loss": 0.031, + "step": 1100 + }, + { + "epoch": 1.8592964824120601, + "grad_norm": 0.1376485973596573, + "learning_rate": 2.9852675602328644e-05, + "loss": 0.0296, + "step": 1110 + }, + { + "epoch": 1.8760469011725294, + "grad_norm": 0.12675292789936066, + "learning_rate": 2.9846839928174923e-05, + "loss": 0.0289, + "step": 1120 + }, + { + "epoch": 1.8927973199329984, + "grad_norm": 0.14450091123580933, + "learning_rate": 2.984089150214065e-05, + "loss": 0.0297, + "step": 1130 + }, + { + "epoch": 1.9095477386934674, + "grad_norm": 0.15321670472621918, + "learning_rate": 2.9834830369400165e-05, + "loss": 0.0295, + "step": 1140 + }, + { + "epoch": 1.9262981574539364, + "grad_norm": 0.13381171226501465, + "learning_rate": 2.9828656575983747e-05, + "loss": 0.0279, + "step": 1150 + }, + { + "epoch": 1.9430485762144054, + "grad_norm": 0.20773489773273468, + "learning_rate": 2.9822370168777253e-05, + "loss": 0.0305, + "step": 1160 + }, + { + "epoch": 1.9597989949748744, + "grad_norm": 0.1649148315191269, + "learning_rate": 2.981597119552177e-05, + "loss": 0.0318, + "step": 1170 + }, + { + "epoch": 1.9765494137353434, + "grad_norm": 0.15318723022937775, + "learning_rate": 2.9809459704813236e-05, + "loss": 0.0283, + "step": 1180 + }, + { + "epoch": 1.9932998324958124, + "grad_norm": 0.11235921084880829, + "learning_rate": 2.980283574610211e-05, + "loss": 0.0281, + "step": 1190 + }, + { + "epoch": 2.0100502512562812, + "grad_norm": 0.14214928448200226, + "learning_rate": 2.9796099369692946e-05, + "loss": 0.0277, + "step": 1200 + }, + { + "epoch": 2.0268006700167502, + "grad_norm": 0.1358565390110016, + "learning_rate": 2.978925062674404e-05, + "loss": 0.0281, + "step": 1210 + }, + { + "epoch": 2.0435510887772192, + "grad_norm": 0.1451619565486908, + "learning_rate": 2.9782289569267057e-05, + "loss": 0.0279, + "step": 1220 + }, + { + "epoch": 2.0603015075376883, + "grad_norm": 0.1291060596704483, + "learning_rate": 2.977521625012659e-05, + "loss": 0.0307, + "step": 1230 + }, + { + "epoch": 2.0770519262981573, + "grad_norm": 0.13382935523986816, + "learning_rate": 2.9768030723039798e-05, + "loss": 0.0281, + "step": 1240 + }, + { + "epoch": 2.0938023450586263, + "grad_norm": 0.1389288753271103, + "learning_rate": 2.9760733042575984e-05, + "loss": 0.0286, + "step": 1250 + }, + { + "epoch": 2.1105527638190953, + "grad_norm": 0.13549239933490753, + "learning_rate": 2.9753323264156184e-05, + "loss": 0.0283, + "step": 1260 + }, + { + "epoch": 2.1273031825795643, + "grad_norm": 0.13530868291854858, + "learning_rate": 2.974580144405273e-05, + "loss": 0.0281, + "step": 1270 + }, + { + "epoch": 2.1440536013400333, + "grad_norm": 0.12069135159254074, + "learning_rate": 2.973816763938886e-05, + "loss": 0.0308, + "step": 1280 + }, + { + "epoch": 2.1608040201005023, + "grad_norm": 0.15732339024543762, + "learning_rate": 2.9730421908138236e-05, + "loss": 0.0274, + "step": 1290 + }, + { + "epoch": 2.1775544388609713, + "grad_norm": 0.14720749855041504, + "learning_rate": 2.9722564309124546e-05, + "loss": 0.0284, + "step": 1300 + }, + { + "epoch": 2.1943048576214403, + "grad_norm": 0.13191711902618408, + "learning_rate": 2.971459490202102e-05, + "loss": 0.0259, + "step": 1310 + }, + { + "epoch": 2.2110552763819094, + "grad_norm": 0.12703394889831543, + "learning_rate": 2.9706513747350017e-05, + "loss": 0.0245, + "step": 1320 + }, + { + "epoch": 2.2278056951423784, + "grad_norm": 0.13217298686504364, + "learning_rate": 2.9698320906482535e-05, + "loss": 0.0307, + "step": 1330 + }, + { + "epoch": 2.2445561139028474, + "grad_norm": 0.12877672910690308, + "learning_rate": 2.9690016441637747e-05, + "loss": 0.0276, + "step": 1340 + }, + { + "epoch": 2.2613065326633164, + "grad_norm": 0.13202929496765137, + "learning_rate": 2.9681600415882556e-05, + "loss": 0.0289, + "step": 1350 + }, + { + "epoch": 2.2780569514237854, + "grad_norm": 0.11203215271234512, + "learning_rate": 2.9673072893131068e-05, + "loss": 0.0274, + "step": 1360 + }, + { + "epoch": 2.2948073701842544, + "grad_norm": 0.09884139150381088, + "learning_rate": 2.9664433938144162e-05, + "loss": 0.0275, + "step": 1370 + }, + { + "epoch": 2.3115577889447234, + "grad_norm": 0.12471596896648407, + "learning_rate": 2.9655683616528958e-05, + "loss": 0.0252, + "step": 1380 + }, + { + "epoch": 2.3283082077051924, + "grad_norm": 0.15619947016239166, + "learning_rate": 2.9646821994738322e-05, + "loss": 0.0267, + "step": 1390 + }, + { + "epoch": 2.3450586264656614, + "grad_norm": 0.11800948530435562, + "learning_rate": 2.9637849140070393e-05, + "loss": 0.025, + "step": 1400 + }, + { + "epoch": 2.3618090452261304, + "grad_norm": 0.12330026179552078, + "learning_rate": 2.9628765120668033e-05, + "loss": 0.0242, + "step": 1410 + }, + { + "epoch": 2.3785594639865995, + "grad_norm": 0.14527784287929535, + "learning_rate": 2.9619570005518336e-05, + "loss": 0.0258, + "step": 1420 + }, + { + "epoch": 2.3953098827470685, + "grad_norm": 0.1384161412715912, + "learning_rate": 2.9610263864452077e-05, + "loss": 0.0245, + "step": 1430 + }, + { + "epoch": 2.4120603015075375, + "grad_norm": 0.14128261804580688, + "learning_rate": 2.9600846768143225e-05, + "loss": 0.0243, + "step": 1440 + }, + { + "epoch": 2.4288107202680065, + "grad_norm": 0.12654323875904083, + "learning_rate": 2.959131878810836e-05, + "loss": 0.0257, + "step": 1450 + }, + { + "epoch": 2.4455611390284755, + "grad_norm": 0.19983522593975067, + "learning_rate": 2.9581679996706157e-05, + "loss": 0.0278, + "step": 1460 + }, + { + "epoch": 2.4623115577889445, + "grad_norm": 0.14568129181861877, + "learning_rate": 2.9571930467136833e-05, + "loss": 0.0256, + "step": 1470 + }, + { + "epoch": 2.4790619765494135, + "grad_norm": 0.12198968231678009, + "learning_rate": 2.956207027344159e-05, + "loss": 0.0257, + "step": 1480 + }, + { + "epoch": 2.4958123953098825, + "grad_norm": 0.11267124861478806, + "learning_rate": 2.9552099490502026e-05, + "loss": 0.0246, + "step": 1490 + }, + { + "epoch": 2.5125628140703515, + "grad_norm": 0.11769300699234009, + "learning_rate": 2.9542018194039634e-05, + "loss": 0.0249, + "step": 1500 + }, + { + "epoch": 2.5293132328308205, + "grad_norm": 0.11569931358098984, + "learning_rate": 2.953182646061514e-05, + "loss": 0.0278, + "step": 1510 + }, + { + "epoch": 2.5460636515912896, + "grad_norm": 0.12251350283622742, + "learning_rate": 2.9521524367627994e-05, + "loss": 0.0233, + "step": 1520 + }, + { + "epoch": 2.5628140703517586, + "grad_norm": 0.11296738684177399, + "learning_rate": 2.9511111993315743e-05, + "loss": 0.0251, + "step": 1530 + }, + { + "epoch": 2.5795644891122276, + "grad_norm": 0.11284096539020538, + "learning_rate": 2.9500589416753447e-05, + "loss": 0.0261, + "step": 1540 + }, + { + "epoch": 2.5963149078726966, + "grad_norm": 0.14394700527191162, + "learning_rate": 2.9489956717853088e-05, + "loss": 0.0272, + "step": 1550 + }, + { + "epoch": 2.6130653266331656, + "grad_norm": 0.12496906518936157, + "learning_rate": 2.9479213977362935e-05, + "loss": 0.0255, + "step": 1560 + }, + { + "epoch": 2.6298157453936346, + "grad_norm": 0.1484213024377823, + "learning_rate": 2.9468361276866973e-05, + "loss": 0.0257, + "step": 1570 + }, + { + "epoch": 2.6465661641541036, + "grad_norm": 0.13880600035190582, + "learning_rate": 2.9457398698784243e-05, + "loss": 0.0226, + "step": 1580 + }, + { + "epoch": 2.6633165829145726, + "grad_norm": 0.1260875016450882, + "learning_rate": 2.944632632636823e-05, + "loss": 0.026, + "step": 1590 + }, + { + "epoch": 2.6800670016750416, + "grad_norm": 0.14514032006263733, + "learning_rate": 2.9435144243706255e-05, + "loss": 0.0251, + "step": 1600 + }, + { + "epoch": 2.6968174204355106, + "grad_norm": 0.10520205646753311, + "learning_rate": 2.9423852535718805e-05, + "loss": 0.023, + "step": 1610 + }, + { + "epoch": 2.7135678391959797, + "grad_norm": 0.09249555319547653, + "learning_rate": 2.9412451288158883e-05, + "loss": 0.0238, + "step": 1620 + }, + { + "epoch": 2.7303182579564487, + "grad_norm": 0.08644948154687881, + "learning_rate": 2.94009405876114e-05, + "loss": 0.0242, + "step": 1630 + }, + { + "epoch": 2.7470686767169177, + "grad_norm": 0.12189530581235886, + "learning_rate": 2.9389320521492464e-05, + "loss": 0.0216, + "step": 1640 + }, + { + "epoch": 2.7638190954773867, + "grad_norm": 0.10813789814710617, + "learning_rate": 2.9377591178048765e-05, + "loss": 0.0249, + "step": 1650 + }, + { + "epoch": 2.7805695142378557, + "grad_norm": 0.10432349145412445, + "learning_rate": 2.9365752646356857e-05, + "loss": 0.0216, + "step": 1660 + }, + { + "epoch": 2.7973199329983247, + "grad_norm": 0.08932849764823914, + "learning_rate": 2.935380501632253e-05, + "loss": 0.0245, + "step": 1670 + }, + { + "epoch": 2.8140703517587937, + "grad_norm": 0.12938445806503296, + "learning_rate": 2.934174837868008e-05, + "loss": 0.0252, + "step": 1680 + }, + { + "epoch": 2.8308207705192627, + "grad_norm": 0.20361070334911346, + "learning_rate": 2.932958282499167e-05, + "loss": 0.0258, + "step": 1690 + }, + { + "epoch": 2.8475711892797317, + "grad_norm": 0.12765756249427795, + "learning_rate": 2.9317308447646585e-05, + "loss": 0.0256, + "step": 1700 + }, + { + "epoch": 2.8643216080402008, + "grad_norm": 0.13113832473754883, + "learning_rate": 2.9304925339860565e-05, + "loss": 0.0258, + "step": 1710 + }, + { + "epoch": 2.8810720268006698, + "grad_norm": 0.13574381172657013, + "learning_rate": 2.929243359567509e-05, + "loss": 0.0231, + "step": 1720 + }, + { + "epoch": 2.8978224455611388, + "grad_norm": 0.15551957488059998, + "learning_rate": 2.9279833309956637e-05, + "loss": 0.0233, + "step": 1730 + }, + { + "epoch": 2.914572864321608, + "grad_norm": 0.1148996576666832, + "learning_rate": 2.9267124578396013e-05, + "loss": 0.0215, + "step": 1740 + }, + { + "epoch": 2.931323283082077, + "grad_norm": 0.1132749691605568, + "learning_rate": 2.9254307497507584e-05, + "loss": 0.0255, + "step": 1750 + }, + { + "epoch": 2.948073701842546, + "grad_norm": 0.12405268102884293, + "learning_rate": 2.924138216462856e-05, + "loss": 0.0232, + "step": 1760 + }, + { + "epoch": 2.964824120603015, + "grad_norm": 0.13782398402690887, + "learning_rate": 2.922834867791825e-05, + "loss": 0.0257, + "step": 1770 + }, + { + "epoch": 2.981574539363484, + "grad_norm": 0.13731901347637177, + "learning_rate": 2.9215207136357322e-05, + "loss": 0.0235, + "step": 1780 + }, + { + "epoch": 2.998324958123953, + "grad_norm": 0.11207369714975357, + "learning_rate": 2.9201957639747056e-05, + "loss": 0.0254, + "step": 1790 + }, + { + "epoch": 3.0150753768844223, + "grad_norm": 0.12427914887666702, + "learning_rate": 2.9188600288708565e-05, + "loss": 0.0231, + "step": 1800 + }, + { + "epoch": 3.0318257956448913, + "grad_norm": 0.12067142128944397, + "learning_rate": 2.9175135184682053e-05, + "loss": 0.0211, + "step": 1810 + }, + { + "epoch": 3.0485762144053603, + "grad_norm": 0.1151166558265686, + "learning_rate": 2.9161562429926038e-05, + "loss": 0.0228, + "step": 1820 + }, + { + "epoch": 3.0653266331658293, + "grad_norm": 0.12743020057678223, + "learning_rate": 2.9147882127516568e-05, + "loss": 0.0227, + "step": 1830 + }, + { + "epoch": 3.0820770519262983, + "grad_norm": 0.12181981652975082, + "learning_rate": 2.913409438134644e-05, + "loss": 0.024, + "step": 1840 + }, + { + "epoch": 3.0988274706867673, + "grad_norm": 0.1195250153541565, + "learning_rate": 2.9120199296124436e-05, + "loss": 0.0251, + "step": 1850 + }, + { + "epoch": 3.1155778894472363, + "grad_norm": 0.09719762206077576, + "learning_rate": 2.9106196977374477e-05, + "loss": 0.0232, + "step": 1860 + }, + { + "epoch": 3.1323283082077054, + "grad_norm": 0.11931390315294266, + "learning_rate": 2.9092087531434878e-05, + "loss": 0.0236, + "step": 1870 + }, + { + "epoch": 3.1490787269681744, + "grad_norm": 0.10895305871963501, + "learning_rate": 2.9077871065457495e-05, + "loss": 0.023, + "step": 1880 + }, + { + "epoch": 3.1658291457286434, + "grad_norm": 0.1338292807340622, + "learning_rate": 2.9063547687406945e-05, + "loss": 0.024, + "step": 1890 + }, + { + "epoch": 3.1825795644891124, + "grad_norm": 0.11511155962944031, + "learning_rate": 2.9049117506059756e-05, + "loss": 0.0208, + "step": 1900 + }, + { + "epoch": 3.1993299832495814, + "grad_norm": 0.11553452163934708, + "learning_rate": 2.9034580631003564e-05, + "loss": 0.0232, + "step": 1910 + }, + { + "epoch": 3.2160804020100504, + "grad_norm": 0.1008048877120018, + "learning_rate": 2.9019937172636276e-05, + "loss": 0.0222, + "step": 1920 + }, + { + "epoch": 3.2328308207705194, + "grad_norm": 0.10765612125396729, + "learning_rate": 2.9005187242165228e-05, + "loss": 0.0226, + "step": 1930 + }, + { + "epoch": 3.2495812395309884, + "grad_norm": 0.11185599863529205, + "learning_rate": 2.899033095160633e-05, + "loss": 0.0213, + "step": 1940 + }, + { + "epoch": 3.2663316582914574, + "grad_norm": 0.09481639415025711, + "learning_rate": 2.8975368413783235e-05, + "loss": 0.0217, + "step": 1950 + }, + { + "epoch": 3.2830820770519265, + "grad_norm": 0.11557790637016296, + "learning_rate": 2.8960299742326473e-05, + "loss": 0.0231, + "step": 1960 + }, + { + "epoch": 3.2998324958123955, + "grad_norm": 0.13437284529209137, + "learning_rate": 2.8945125051672595e-05, + "loss": 0.0234, + "step": 1970 + }, + { + "epoch": 3.3165829145728645, + "grad_norm": 0.11012782156467438, + "learning_rate": 2.892984445706328e-05, + "loss": 0.0225, + "step": 1980 + }, + { + "epoch": 3.3333333333333335, + "grad_norm": 0.11317654699087143, + "learning_rate": 2.891445807454449e-05, + "loss": 0.024, + "step": 1990 + }, + { + "epoch": 3.3500837520938025, + "grad_norm": 0.19488593935966492, + "learning_rate": 2.8898966020965576e-05, + "loss": 0.024, + "step": 2000 + }, + { + "epoch": 3.3668341708542715, + "grad_norm": 0.10659220814704895, + "learning_rate": 2.8883368413978386e-05, + "loss": 0.0225, + "step": 2010 + }, + { + "epoch": 3.3835845896147405, + "grad_norm": 0.09751500934362411, + "learning_rate": 2.8867665372036375e-05, + "loss": 0.0258, + "step": 2020 + }, + { + "epoch": 3.4003350083752095, + "grad_norm": 0.09632577747106552, + "learning_rate": 2.8851857014393705e-05, + "loss": 0.0222, + "step": 2030 + }, + { + "epoch": 3.4170854271356785, + "grad_norm": 0.12793496251106262, + "learning_rate": 2.8835943461104354e-05, + "loss": 0.0214, + "step": 2040 + }, + { + "epoch": 3.4338358458961475, + "grad_norm": 0.11942226439714432, + "learning_rate": 2.8819924833021176e-05, + "loss": 0.023, + "step": 2050 + }, + { + "epoch": 3.4505862646566166, + "grad_norm": 0.10179294645786285, + "learning_rate": 2.8803801251795007e-05, + "loss": 0.0212, + "step": 2060 + }, + { + "epoch": 3.4673366834170856, + "grad_norm": 0.1228783056139946, + "learning_rate": 2.8787572839873725e-05, + "loss": 0.0225, + "step": 2070 + }, + { + "epoch": 3.4840871021775546, + "grad_norm": 0.13721901178359985, + "learning_rate": 2.8771239720501333e-05, + "loss": 0.0199, + "step": 2080 + }, + { + "epoch": 3.5008375209380236, + "grad_norm": 0.1087748259305954, + "learning_rate": 2.8754802017717015e-05, + "loss": 0.0228, + "step": 2090 + }, + { + "epoch": 3.5175879396984926, + "grad_norm": 0.14415811002254486, + "learning_rate": 2.87382598563542e-05, + "loss": 0.0241, + "step": 2100 + }, + { + "epoch": 3.5343383584589616, + "grad_norm": 0.10477285832166672, + "learning_rate": 2.8721613362039607e-05, + "loss": 0.02, + "step": 2110 + }, + { + "epoch": 3.5510887772194306, + "grad_norm": 0.10786721855401993, + "learning_rate": 2.8704862661192295e-05, + "loss": 0.0212, + "step": 2120 + }, + { + "epoch": 3.5678391959798996, + "grad_norm": 0.10725842416286469, + "learning_rate": 2.86880078810227e-05, + "loss": 0.022, + "step": 2130 + }, + { + "epoch": 3.5845896147403686, + "grad_norm": 0.11172711849212646, + "learning_rate": 2.8671049149531677e-05, + "loss": 0.021, + "step": 2140 + }, + { + "epoch": 3.6013400335008376, + "grad_norm": 0.10756686329841614, + "learning_rate": 2.8653986595509524e-05, + "loss": 0.0208, + "step": 2150 + }, + { + "epoch": 3.6180904522613067, + "grad_norm": 0.08842871338129044, + "learning_rate": 2.8636820348534987e-05, + "loss": 0.0208, + "step": 2160 + }, + { + "epoch": 3.6348408710217757, + "grad_norm": 0.10979699343442917, + "learning_rate": 2.861955053897431e-05, + "loss": 0.0204, + "step": 2170 + }, + { + "epoch": 3.6515912897822447, + "grad_norm": 0.0892506092786789, + "learning_rate": 2.8602177297980213e-05, + "loss": 0.0209, + "step": 2180 + }, + { + "epoch": 3.6683417085427137, + "grad_norm": 0.10864245146512985, + "learning_rate": 2.8584700757490913e-05, + "loss": 0.0214, + "step": 2190 + }, + { + "epoch": 3.6850921273031827, + "grad_norm": 0.12509699165821075, + "learning_rate": 2.8567121050229122e-05, + "loss": 0.0236, + "step": 2200 + }, + { + "epoch": 3.7018425460636517, + "grad_norm": 0.11275574564933777, + "learning_rate": 2.8549438309701028e-05, + "loss": 0.0237, + "step": 2210 + }, + { + "epoch": 3.7185929648241207, + "grad_norm": 0.12549647688865662, + "learning_rate": 2.8531652670195292e-05, + "loss": 0.021, + "step": 2220 + }, + { + "epoch": 3.7353433835845897, + "grad_norm": 0.11705254763364792, + "learning_rate": 2.8513764266782025e-05, + "loss": 0.0224, + "step": 2230 + }, + { + "epoch": 3.7520938023450587, + "grad_norm": 0.16595874726772308, + "learning_rate": 2.849577323531176e-05, + "loss": 0.0224, + "step": 2240 + }, + { + "epoch": 3.7688442211055277, + "grad_norm": 0.10253427922725677, + "learning_rate": 2.8477679712414424e-05, + "loss": 0.0215, + "step": 2250 + }, + { + "epoch": 3.7855946398659968, + "grad_norm": 0.10289912670850754, + "learning_rate": 2.8459483835498298e-05, + "loss": 0.0237, + "step": 2260 + }, + { + "epoch": 3.8023450586264658, + "grad_norm": 0.11400792002677917, + "learning_rate": 2.844118574274896e-05, + "loss": 0.0205, + "step": 2270 + }, + { + "epoch": 3.819095477386935, + "grad_norm": 0.11081373691558838, + "learning_rate": 2.842278557312828e-05, + "loss": 0.0184, + "step": 2280 + }, + { + "epoch": 3.835845896147404, + "grad_norm": 0.13571597635746002, + "learning_rate": 2.84042834663733e-05, + "loss": 0.0207, + "step": 2290 + }, + { + "epoch": 3.852596314907873, + "grad_norm": 0.10395380854606628, + "learning_rate": 2.8385679562995235e-05, + "loss": 0.0209, + "step": 2300 + }, + { + "epoch": 3.869346733668342, + "grad_norm": 0.10459823906421661, + "learning_rate": 2.8366974004278357e-05, + "loss": 0.0242, + "step": 2310 + }, + { + "epoch": 3.886097152428811, + "grad_norm": 0.10516241937875748, + "learning_rate": 2.834816693227895e-05, + "loss": 0.0205, + "step": 2320 + }, + { + "epoch": 3.90284757118928, + "grad_norm": 0.1241426020860672, + "learning_rate": 2.8329258489824234e-05, + "loss": 0.0227, + "step": 2330 + }, + { + "epoch": 3.919597989949749, + "grad_norm": 0.10768191516399384, + "learning_rate": 2.8310248820511248e-05, + "loss": 0.0215, + "step": 2340 + }, + { + "epoch": 3.936348408710218, + "grad_norm": 0.10944578796625137, + "learning_rate": 2.829113806870581e-05, + "loss": 0.0214, + "step": 2350 + }, + { + "epoch": 3.953098827470687, + "grad_norm": 0.11351463198661804, + "learning_rate": 2.8271926379541363e-05, + "loss": 0.022, + "step": 2360 + }, + { + "epoch": 3.969849246231156, + "grad_norm": 0.12451263517141342, + "learning_rate": 2.8252613898917923e-05, + "loss": 0.0216, + "step": 2370 + }, + { + "epoch": 3.986599664991625, + "grad_norm": 0.11138986051082611, + "learning_rate": 2.8233200773500942e-05, + "loss": 0.0215, + "step": 2380 + }, + { + "epoch": 4.0033500837520934, + "grad_norm": 0.14621508121490479, + "learning_rate": 2.8213687150720204e-05, + "loss": 0.0227, + "step": 2390 + }, + { + "epoch": 4.0201005025125625, + "grad_norm": 0.12678438425064087, + "learning_rate": 2.8194073178768703e-05, + "loss": 0.0211, + "step": 2400 + }, + { + "epoch": 4.0368509212730315, + "grad_norm": 0.10121660679578781, + "learning_rate": 2.817435900660152e-05, + "loss": 0.0191, + "step": 2410 + }, + { + "epoch": 4.0536013400335005, + "grad_norm": 0.09031516313552856, + "learning_rate": 2.8154544783934688e-05, + "loss": 0.0241, + "step": 2420 + }, + { + "epoch": 4.0703517587939695, + "grad_norm": 0.12526579201221466, + "learning_rate": 2.8134630661244055e-05, + "loss": 0.0212, + "step": 2430 + }, + { + "epoch": 4.0871021775544385, + "grad_norm": 0.09737418591976166, + "learning_rate": 2.8114616789764146e-05, + "loss": 0.0196, + "step": 2440 + }, + { + "epoch": 4.1038525963149075, + "grad_norm": 0.12097261101007462, + "learning_rate": 2.8094503321487017e-05, + "loss": 0.0208, + "step": 2450 + }, + { + "epoch": 4.1206030150753765, + "grad_norm": 0.10906875133514404, + "learning_rate": 2.8074290409161084e-05, + "loss": 0.0183, + "step": 2460 + }, + { + "epoch": 4.1373534338358455, + "grad_norm": 0.12879160046577454, + "learning_rate": 2.8053978206289984e-05, + "loss": 0.0194, + "step": 2470 + }, + { + "epoch": 4.1541038525963145, + "grad_norm": 0.0820513367652893, + "learning_rate": 2.803356686713139e-05, + "loss": 0.0214, + "step": 2480 + }, + { + "epoch": 4.1708542713567835, + "grad_norm": 0.07784825563430786, + "learning_rate": 2.801305654669587e-05, + "loss": 0.0217, + "step": 2490 + }, + { + "epoch": 4.187604690117253, + "grad_norm": 0.0847000777721405, + "learning_rate": 2.799244740074566e-05, + "loss": 0.0195, + "step": 2500 + }, + { + "epoch": 4.204355108877722, + "grad_norm": 0.10731722414493561, + "learning_rate": 2.797173958579353e-05, + "loss": 0.0206, + "step": 2510 + }, + { + "epoch": 4.221105527638191, + "grad_norm": 0.1059015691280365, + "learning_rate": 2.7950933259101575e-05, + "loss": 0.0189, + "step": 2520 + }, + { + "epoch": 4.23785594639866, + "grad_norm": 0.10778709501028061, + "learning_rate": 2.793002857868001e-05, + "loss": 0.0208, + "step": 2530 + }, + { + "epoch": 4.254606365159129, + "grad_norm": 0.08753228187561035, + "learning_rate": 2.7909025703285983e-05, + "loss": 0.0195, + "step": 2540 + }, + { + "epoch": 4.271356783919598, + "grad_norm": 0.1175403892993927, + "learning_rate": 2.7887924792422385e-05, + "loss": 0.023, + "step": 2550 + }, + { + "epoch": 4.288107202680067, + "grad_norm": 0.12728635966777802, + "learning_rate": 2.78667260063366e-05, + "loss": 0.0215, + "step": 2560 + }, + { + "epoch": 4.304857621440536, + "grad_norm": 0.10965393483638763, + "learning_rate": 2.784542950601931e-05, + "loss": 0.019, + "step": 2570 + }, + { + "epoch": 4.321608040201005, + "grad_norm": 0.10030016303062439, + "learning_rate": 2.7824035453203293e-05, + "loss": 0.0214, + "step": 2580 + }, + { + "epoch": 4.338358458961474, + "grad_norm": 0.10159602761268616, + "learning_rate": 2.7802544010362154e-05, + "loss": 0.018, + "step": 2590 + }, + { + "epoch": 4.355108877721943, + "grad_norm": 0.08524280786514282, + "learning_rate": 2.778095534070912e-05, + "loss": 0.0225, + "step": 2600 + }, + { + "epoch": 4.371859296482412, + "grad_norm": 0.08864463865756989, + "learning_rate": 2.7759269608195777e-05, + "loss": 0.0214, + "step": 2610 + }, + { + "epoch": 4.388609715242881, + "grad_norm": 0.0836586281657219, + "learning_rate": 2.773748697751086e-05, + "loss": 0.019, + "step": 2620 + }, + { + "epoch": 4.40536013400335, + "grad_norm": 0.10019984096288681, + "learning_rate": 2.7715607614078956e-05, + "loss": 0.0204, + "step": 2630 + }, + { + "epoch": 4.422110552763819, + "grad_norm": 0.097859688103199, + "learning_rate": 2.7693631684059305e-05, + "loss": 0.0217, + "step": 2640 + }, + { + "epoch": 4.438860971524288, + "grad_norm": 0.09158801287412643, + "learning_rate": 2.767155935434448e-05, + "loss": 0.0197, + "step": 2650 + }, + { + "epoch": 4.455611390284757, + "grad_norm": 0.09695200622081757, + "learning_rate": 2.7649390792559146e-05, + "loss": 0.0206, + "step": 2660 + }, + { + "epoch": 4.472361809045226, + "grad_norm": 0.09822610765695572, + "learning_rate": 2.7627126167058807e-05, + "loss": 0.0203, + "step": 2670 + }, + { + "epoch": 4.489112227805695, + "grad_norm": 0.09746448695659637, + "learning_rate": 2.7604765646928482e-05, + "loss": 0.0204, + "step": 2680 + }, + { + "epoch": 4.505862646566165, + "grad_norm": 0.10244615375995636, + "learning_rate": 2.758230940198147e-05, + "loss": 0.0215, + "step": 2690 + }, + { + "epoch": 4.522613065326633, + "grad_norm": 0.10013485699892044, + "learning_rate": 2.755975760275802e-05, + "loss": 0.0198, + "step": 2700 + }, + { + "epoch": 4.539363484087103, + "grad_norm": 0.09892448037862778, + "learning_rate": 2.7537110420524057e-05, + "loss": 0.0199, + "step": 2710 + }, + { + "epoch": 4.556113902847571, + "grad_norm": 0.09825485199689865, + "learning_rate": 2.7514368027269875e-05, + "loss": 0.0186, + "step": 2720 + }, + { + "epoch": 4.572864321608041, + "grad_norm": 0.11656782776117325, + "learning_rate": 2.749153059570883e-05, + "loss": 0.0217, + "step": 2730 + }, + { + "epoch": 4.589614740368509, + "grad_norm": 0.08932335674762726, + "learning_rate": 2.7468598299276044e-05, + "loss": 0.0197, + "step": 2740 + }, + { + "epoch": 4.606365159128979, + "grad_norm": 0.09198352694511414, + "learning_rate": 2.7445571312127058e-05, + "loss": 0.0168, + "step": 2750 + }, + { + "epoch": 4.623115577889447, + "grad_norm": 0.09846565872430801, + "learning_rate": 2.7422449809136538e-05, + "loss": 0.0183, + "step": 2760 + }, + { + "epoch": 4.639865996649917, + "grad_norm": 0.07488419115543365, + "learning_rate": 2.739923396589693e-05, + "loss": 0.0209, + "step": 2770 + }, + { + "epoch": 4.656616415410385, + "grad_norm": 0.11510682106018066, + "learning_rate": 2.737592395871713e-05, + "loss": 0.0184, + "step": 2780 + }, + { + "epoch": 4.673366834170855, + "grad_norm": 0.08309126645326614, + "learning_rate": 2.7352519964621146e-05, + "loss": 0.0175, + "step": 2790 + }, + { + "epoch": 4.690117252931323, + "grad_norm": 0.0989244133234024, + "learning_rate": 2.7329022161346758e-05, + "loss": 0.02, + "step": 2800 + }, + { + "epoch": 4.706867671691793, + "grad_norm": 0.09160429984331131, + "learning_rate": 2.730543072734417e-05, + "loss": 0.0211, + "step": 2810 + }, + { + "epoch": 4.723618090452261, + "grad_norm": 0.13737095892429352, + "learning_rate": 2.728174584177463e-05, + "loss": 0.02, + "step": 2820 + }, + { + "epoch": 4.740368509212731, + "grad_norm": 0.09824742376804352, + "learning_rate": 2.725796768450911e-05, + "loss": 0.0187, + "step": 2830 + }, + { + "epoch": 4.757118927973199, + "grad_norm": 0.07674989849328995, + "learning_rate": 2.7234096436126903e-05, + "loss": 0.0201, + "step": 2840 + }, + { + "epoch": 4.773869346733669, + "grad_norm": 0.09987271577119827, + "learning_rate": 2.721013227791428e-05, + "loss": 0.0201, + "step": 2850 + }, + { + "epoch": 4.790619765494137, + "grad_norm": 0.10610336065292358, + "learning_rate": 2.7186075391863088e-05, + "loss": 0.0216, + "step": 2860 + }, + { + "epoch": 4.807370184254607, + "grad_norm": 0.09385973960161209, + "learning_rate": 2.7161925960669383e-05, + "loss": 0.0218, + "step": 2870 + }, + { + "epoch": 4.824120603015075, + "grad_norm": 0.1261572688817978, + "learning_rate": 2.713768416773205e-05, + "loss": 0.0201, + "step": 2880 + }, + { + "epoch": 4.840871021775545, + "grad_norm": 0.06238450109958649, + "learning_rate": 2.7113350197151394e-05, + "loss": 0.0192, + "step": 2890 + }, + { + "epoch": 4.857621440536013, + "grad_norm": 0.08462442457675934, + "learning_rate": 2.708892423372774e-05, + "loss": 0.0189, + "step": 2900 + }, + { + "epoch": 4.874371859296483, + "grad_norm": 0.11602508276700974, + "learning_rate": 2.7064406462960044e-05, + "loss": 0.0175, + "step": 2910 + }, + { + "epoch": 4.891122278056951, + "grad_norm": 0.08591975271701813, + "learning_rate": 2.7039797071044482e-05, + "loss": 0.023, + "step": 2920 + }, + { + "epoch": 4.907872696817421, + "grad_norm": 0.11215902119874954, + "learning_rate": 2.7015096244873028e-05, + "loss": 0.0204, + "step": 2930 + }, + { + "epoch": 4.924623115577889, + "grad_norm": 0.11343713849782944, + "learning_rate": 2.6990304172032038e-05, + "loss": 0.0199, + "step": 2940 + }, + { + "epoch": 4.941373534338359, + "grad_norm": 0.08644749969244003, + "learning_rate": 2.6965421040800823e-05, + "loss": 0.0219, + "step": 2950 + }, + { + "epoch": 4.958123953098827, + "grad_norm": 0.0927010178565979, + "learning_rate": 2.6940447040150228e-05, + "loss": 0.0194, + "step": 2960 + }, + { + "epoch": 4.974874371859297, + "grad_norm": 0.10158302634954453, + "learning_rate": 2.691538235974118e-05, + "loss": 0.0205, + "step": 2970 + }, + { + "epoch": 4.991624790619765, + "grad_norm": 0.10737495869398117, + "learning_rate": 2.6890227189923285e-05, + "loss": 0.0189, + "step": 2980 + }, + { + "epoch": 5.008375209380235, + "grad_norm": 0.07879546284675598, + "learning_rate": 2.6864981721733317e-05, + "loss": 0.0179, + "step": 2990 + }, + { + "epoch": 5.025125628140704, + "grad_norm": 0.10061459243297577, + "learning_rate": 2.683964614689383e-05, + "loss": 0.0196, + "step": 3000 + }, + { + "epoch": 5.041876046901173, + "grad_norm": 0.05866040289402008, + "learning_rate": 2.681422065781169e-05, + "loss": 0.0156, + "step": 3010 + }, + { + "epoch": 5.058626465661642, + "grad_norm": 0.13595989346504211, + "learning_rate": 2.6788705447576563e-05, + "loss": 0.0185, + "step": 3020 + }, + { + "epoch": 5.075376884422111, + "grad_norm": 0.10418418049812317, + "learning_rate": 2.6763100709959524e-05, + "loss": 0.0172, + "step": 3030 + }, + { + "epoch": 5.09212730318258, + "grad_norm": 0.09122584760189056, + "learning_rate": 2.6737406639411537e-05, + "loss": 0.0193, + "step": 3040 + }, + { + "epoch": 5.108877721943049, + "grad_norm": 0.10324976593255997, + "learning_rate": 2.6711623431061985e-05, + "loss": 0.0171, + "step": 3050 + }, + { + "epoch": 5.125628140703517, + "grad_norm": 0.11300911009311676, + "learning_rate": 2.6685751280717195e-05, + "loss": 0.0176, + "step": 3060 + }, + { + "epoch": 5.142378559463987, + "grad_norm": 0.08261276036500931, + "learning_rate": 2.665979038485896e-05, + "loss": 0.0195, + "step": 3070 + }, + { + "epoch": 5.159128978224456, + "grad_norm": 0.10486511141061783, + "learning_rate": 2.6633740940643022e-05, + "loss": 0.017, + "step": 3080 + }, + { + "epoch": 5.175879396984925, + "grad_norm": 0.10391178727149963, + "learning_rate": 2.6607603145897594e-05, + "loss": 0.0201, + "step": 3090 + }, + { + "epoch": 5.192629815745394, + "grad_norm": 0.0923885777592659, + "learning_rate": 2.6581377199121853e-05, + "loss": 0.0191, + "step": 3100 + }, + { + "epoch": 5.209380234505863, + "grad_norm": 0.12680231034755707, + "learning_rate": 2.6555063299484434e-05, + "loss": 0.0177, + "step": 3110 + }, + { + "epoch": 5.226130653266332, + "grad_norm": 0.07475026696920395, + "learning_rate": 2.6528661646821917e-05, + "loss": 0.0193, + "step": 3120 + }, + { + "epoch": 5.242881072026801, + "grad_norm": 0.10191482305526733, + "learning_rate": 2.6502172441637302e-05, + "loss": 0.0193, + "step": 3130 + }, + { + "epoch": 5.259631490787269, + "grad_norm": 0.08599104732275009, + "learning_rate": 2.6475595885098493e-05, + "loss": 0.0193, + "step": 3140 + }, + { + "epoch": 5.276381909547739, + "grad_norm": 0.0979737862944603, + "learning_rate": 2.644893217903677e-05, + "loss": 0.0177, + "step": 3150 + }, + { + "epoch": 5.293132328308207, + "grad_norm": 0.1444755345582962, + "learning_rate": 2.6422181525945265e-05, + "loss": 0.0185, + "step": 3160 + }, + { + "epoch": 5.309882747068677, + "grad_norm": 0.0918390303850174, + "learning_rate": 2.63953441289774e-05, + "loss": 0.02, + "step": 3170 + }, + { + "epoch": 5.326633165829146, + "grad_norm": 0.11050040274858475, + "learning_rate": 2.6368420191945365e-05, + "loss": 0.0181, + "step": 3180 + }, + { + "epoch": 5.343383584589615, + "grad_norm": 0.0900280773639679, + "learning_rate": 2.6341409919318572e-05, + "loss": 0.017, + "step": 3190 + }, + { + "epoch": 5.360134003350084, + "grad_norm": 0.08904273808002472, + "learning_rate": 2.6314313516222084e-05, + "loss": 0.0195, + "step": 3200 + }, + { + "epoch": 5.376884422110553, + "grad_norm": 0.10394366830587387, + "learning_rate": 2.6287131188435075e-05, + "loss": 0.0193, + "step": 3210 + }, + { + "epoch": 5.393634840871022, + "grad_norm": 0.10032975673675537, + "learning_rate": 2.6259863142389252e-05, + "loss": 0.0164, + "step": 3220 + }, + { + "epoch": 5.410385259631491, + "grad_norm": 0.09366985410451889, + "learning_rate": 2.62325095851673e-05, + "loss": 0.0171, + "step": 3230 + }, + { + "epoch": 5.42713567839196, + "grad_norm": 0.08237487822771072, + "learning_rate": 2.620507072450131e-05, + "loss": 0.0191, + "step": 3240 + }, + { + "epoch": 5.443886097152429, + "grad_norm": 0.10653271526098251, + "learning_rate": 2.6177546768771183e-05, + "loss": 0.0173, + "step": 3250 + }, + { + "epoch": 5.460636515912898, + "grad_norm": 0.08800845593214035, + "learning_rate": 2.6149937927003062e-05, + "loss": 0.018, + "step": 3260 + }, + { + "epoch": 5.477386934673367, + "grad_norm": 0.10618974268436432, + "learning_rate": 2.6122244408867766e-05, + "loss": 0.0198, + "step": 3270 + }, + { + "epoch": 5.494137353433836, + "grad_norm": 0.0838366374373436, + "learning_rate": 2.6094466424679148e-05, + "loss": 0.0196, + "step": 3280 + }, + { + "epoch": 5.510887772194305, + "grad_norm": 0.08228679746389389, + "learning_rate": 2.606660418539253e-05, + "loss": 0.0208, + "step": 3290 + }, + { + "epoch": 5.527638190954773, + "grad_norm": 0.10232897847890854, + "learning_rate": 2.6038657902603107e-05, + "loss": 0.0173, + "step": 3300 + }, + { + "epoch": 5.544388609715243, + "grad_norm": 0.0675511434674263, + "learning_rate": 2.6010627788544314e-05, + "loss": 0.0171, + "step": 3310 + }, + { + "epoch": 5.561139028475711, + "grad_norm": 0.11169396340847015, + "learning_rate": 2.5982514056086238e-05, + "loss": 0.0202, + "step": 3320 + }, + { + "epoch": 5.577889447236181, + "grad_norm": 0.09199520200490952, + "learning_rate": 2.5954316918733992e-05, + "loss": 0.0188, + "step": 3330 + }, + { + "epoch": 5.594639865996649, + "grad_norm": 0.09636520594358444, + "learning_rate": 2.5926036590626093e-05, + "loss": 0.0189, + "step": 3340 + }, + { + "epoch": 5.611390284757119, + "grad_norm": 0.1068258136510849, + "learning_rate": 2.589767328653283e-05, + "loss": 0.0185, + "step": 3350 + }, + { + "epoch": 5.628140703517588, + "grad_norm": 0.0738004669547081, + "learning_rate": 2.5869227221854646e-05, + "loss": 0.0153, + "step": 3360 + }, + { + "epoch": 5.644891122278057, + "grad_norm": 0.08421333134174347, + "learning_rate": 2.584069861262049e-05, + "loss": 0.0199, + "step": 3370 + }, + { + "epoch": 5.661641541038526, + "grad_norm": 0.1111551895737648, + "learning_rate": 2.58120876754862e-05, + "loss": 0.0147, + "step": 3380 + }, + { + "epoch": 5.678391959798995, + "grad_norm": 0.0835203304886818, + "learning_rate": 2.5783394627732815e-05, + "loss": 0.0168, + "step": 3390 + }, + { + "epoch": 5.695142378559464, + "grad_norm": 0.12761108577251434, + "learning_rate": 2.5754619687264957e-05, + "loss": 0.0189, + "step": 3400 + }, + { + "epoch": 5.711892797319933, + "grad_norm": 0.0886048823595047, + "learning_rate": 2.5725763072609178e-05, + "loss": 0.0164, + "step": 3410 + }, + { + "epoch": 5.728643216080402, + "grad_norm": 0.09086675196886063, + "learning_rate": 2.569682500291229e-05, + "loss": 0.0151, + "step": 3420 + }, + { + "epoch": 5.745393634840871, + "grad_norm": 0.10654979199171066, + "learning_rate": 2.5667805697939687e-05, + "loss": 0.0173, + "step": 3430 + }, + { + "epoch": 5.76214405360134, + "grad_norm": 0.09250108897686005, + "learning_rate": 2.5638705378073704e-05, + "loss": 0.0207, + "step": 3440 + }, + { + "epoch": 5.778894472361809, + "grad_norm": 0.09105037152767181, + "learning_rate": 2.5609524264311934e-05, + "loss": 0.0175, + "step": 3450 + }, + { + "epoch": 5.795644891122278, + "grad_norm": 0.10493557155132294, + "learning_rate": 2.5580262578265536e-05, + "loss": 0.0217, + "step": 3460 + }, + { + "epoch": 5.812395309882747, + "grad_norm": 0.09852892905473709, + "learning_rate": 2.5550920542157557e-05, + "loss": 0.0153, + "step": 3470 + }, + { + "epoch": 5.8291457286432165, + "grad_norm": 0.09802744537591934, + "learning_rate": 2.5521498378821273e-05, + "loss": 0.0195, + "step": 3480 + }, + { + "epoch": 5.8458961474036855, + "grad_norm": 0.07921155542135239, + "learning_rate": 2.549199631169844e-05, + "loss": 0.0171, + "step": 3490 + }, + { + "epoch": 5.8626465661641545, + "grad_norm": 0.09621422737836838, + "learning_rate": 2.5462414564837663e-05, + "loss": 0.0168, + "step": 3500 + }, + { + "epoch": 5.8793969849246235, + "grad_norm": 0.0989849641919136, + "learning_rate": 2.543275336289264e-05, + "loss": 0.017, + "step": 3510 + }, + { + "epoch": 5.8961474036850925, + "grad_norm": 0.11342565715312958, + "learning_rate": 2.540301293112048e-05, + "loss": 0.018, + "step": 3520 + }, + { + "epoch": 5.9128978224455615, + "grad_norm": 0.1399437040090561, + "learning_rate": 2.537319349538e-05, + "loss": 0.0182, + "step": 3530 + }, + { + "epoch": 5.9296482412060305, + "grad_norm": 0.09442800283432007, + "learning_rate": 2.534329528212999e-05, + "loss": 0.0164, + "step": 3540 + }, + { + "epoch": 5.9463986599664995, + "grad_norm": 0.09357256442308426, + "learning_rate": 2.531331851842751e-05, + "loss": 0.017, + "step": 3550 + }, + { + "epoch": 5.9631490787269685, + "grad_norm": 0.10102587193250656, + "learning_rate": 2.528326343192615e-05, + "loss": 0.0161, + "step": 3560 + }, + { + "epoch": 5.9798994974874375, + "grad_norm": 0.09273260831832886, + "learning_rate": 2.5253130250874316e-05, + "loss": 0.0168, + "step": 3570 + }, + { + "epoch": 5.9966499162479066, + "grad_norm": 0.09276672452688217, + "learning_rate": 2.5222919204113494e-05, + "loss": 0.0198, + "step": 3580 + }, + { + "epoch": 6.013400335008376, + "grad_norm": 0.15660251677036285, + "learning_rate": 2.5192630521076494e-05, + "loss": 0.0171, + "step": 3590 + }, + { + "epoch": 6.030150753768845, + "grad_norm": 0.10578886419534683, + "learning_rate": 2.5162264431785734e-05, + "loss": 0.0199, + "step": 3600 + }, + { + "epoch": 6.046901172529314, + "grad_norm": 0.10759575664997101, + "learning_rate": 2.5131821166851484e-05, + "loss": 0.0167, + "step": 3610 + }, + { + "epoch": 6.063651591289783, + "grad_norm": 0.08797171711921692, + "learning_rate": 2.5101300957470095e-05, + "loss": 0.0177, + "step": 3620 + }, + { + "epoch": 6.080402010050252, + "grad_norm": 0.06793040037155151, + "learning_rate": 2.507070403542228e-05, + "loss": 0.0166, + "step": 3630 + }, + { + "epoch": 6.097152428810721, + "grad_norm": 0.08354511857032776, + "learning_rate": 2.504003063307132e-05, + "loss": 0.0177, + "step": 3640 + }, + { + "epoch": 6.11390284757119, + "grad_norm": 0.06704747676849365, + "learning_rate": 2.500928098336131e-05, + "loss": 0.0184, + "step": 3650 + }, + { + "epoch": 6.130653266331659, + "grad_norm": 0.122153140604496, + "learning_rate": 2.497845531981541e-05, + "loss": 0.0191, + "step": 3660 + }, + { + "epoch": 6.147403685092128, + "grad_norm": 0.08629975467920303, + "learning_rate": 2.4947553876534047e-05, + "loss": 0.0175, + "step": 3670 + }, + { + "epoch": 6.164154103852597, + "grad_norm": 0.09296640753746033, + "learning_rate": 2.4916576888193137e-05, + "loss": 0.0148, + "step": 3680 + }, + { + "epoch": 6.180904522613066, + "grad_norm": 0.06364050507545471, + "learning_rate": 2.488552459004231e-05, + "loss": 0.0169, + "step": 3690 + }, + { + "epoch": 6.197654941373535, + "grad_norm": 0.07617541402578354, + "learning_rate": 2.4854397217903145e-05, + "loss": 0.0176, + "step": 3700 + }, + { + "epoch": 6.214405360134004, + "grad_norm": 0.08891832083463669, + "learning_rate": 2.4823195008167325e-05, + "loss": 0.0172, + "step": 3710 + }, + { + "epoch": 6.231155778894473, + "grad_norm": 0.06869089603424072, + "learning_rate": 2.4791918197794907e-05, + "loss": 0.0161, + "step": 3720 + }, + { + "epoch": 6.247906197654942, + "grad_norm": 0.10788177698850632, + "learning_rate": 2.4760567024312476e-05, + "loss": 0.0154, + "step": 3730 + }, + { + "epoch": 6.264656616415411, + "grad_norm": 0.14439617097377777, + "learning_rate": 2.4729141725811344e-05, + "loss": 0.0183, + "step": 3740 + }, + { + "epoch": 6.28140703517588, + "grad_norm": 0.10388529300689697, + "learning_rate": 2.4697642540945774e-05, + "loss": 0.0194, + "step": 3750 + }, + { + "epoch": 6.298157453936349, + "grad_norm": 0.08002973347902298, + "learning_rate": 2.4666069708931136e-05, + "loss": 0.015, + "step": 3760 + }, + { + "epoch": 6.314907872696818, + "grad_norm": 0.0883835032582283, + "learning_rate": 2.46344234695421e-05, + "loss": 0.0183, + "step": 3770 + }, + { + "epoch": 6.331658291457287, + "grad_norm": 0.08702576160430908, + "learning_rate": 2.460270406311082e-05, + "loss": 0.0159, + "step": 3780 + }, + { + "epoch": 6.348408710217756, + "grad_norm": 0.09502051025629044, + "learning_rate": 2.457091173052511e-05, + "loss": 0.0173, + "step": 3790 + }, + { + "epoch": 6.365159128978225, + "grad_norm": 0.09019004553556442, + "learning_rate": 2.4539046713226593e-05, + "loss": 0.0192, + "step": 3800 + }, + { + "epoch": 6.381909547738694, + "grad_norm": 0.10312219709157944, + "learning_rate": 2.4507109253208898e-05, + "loss": 0.0154, + "step": 3810 + }, + { + "epoch": 6.398659966499163, + "grad_norm": 0.08851608633995056, + "learning_rate": 2.4475099593015813e-05, + "loss": 0.0172, + "step": 3820 + }, + { + "epoch": 6.415410385259632, + "grad_norm": 0.0891028419137001, + "learning_rate": 2.4443017975739423e-05, + "loss": 0.0188, + "step": 3830 + }, + { + "epoch": 6.432160804020101, + "grad_norm": 0.08572867512702942, + "learning_rate": 2.4410864645018288e-05, + "loss": 0.0165, + "step": 3840 + }, + { + "epoch": 6.44891122278057, + "grad_norm": 0.09533847123384476, + "learning_rate": 2.4378639845035583e-05, + "loss": 0.0158, + "step": 3850 + }, + { + "epoch": 6.465661641541039, + "grad_norm": 0.0869334265589714, + "learning_rate": 2.4346343820517237e-05, + "loss": 0.0158, + "step": 3860 + }, + { + "epoch": 6.482412060301508, + "grad_norm": 0.10855986922979355, + "learning_rate": 2.4313976816730095e-05, + "loss": 0.0179, + "step": 3870 + }, + { + "epoch": 6.499162479061977, + "grad_norm": 0.0890968069434166, + "learning_rate": 2.4281539079480047e-05, + "loss": 0.0166, + "step": 3880 + }, + { + "epoch": 6.515912897822446, + "grad_norm": 0.09087546169757843, + "learning_rate": 2.4249030855110128e-05, + "loss": 0.0156, + "step": 3890 + }, + { + "epoch": 6.532663316582915, + "grad_norm": 0.07741452008485794, + "learning_rate": 2.42164523904987e-05, + "loss": 0.0163, + "step": 3900 + }, + { + "epoch": 6.549413735343384, + "grad_norm": 0.09288477897644043, + "learning_rate": 2.4183803933057554e-05, + "loss": 0.0165, + "step": 3910 + }, + { + "epoch": 6.566164154103853, + "grad_norm": 0.08580487966537476, + "learning_rate": 2.4151085730730013e-05, + "loss": 0.0201, + "step": 3920 + }, + { + "epoch": 6.582914572864322, + "grad_norm": 0.079682357609272, + "learning_rate": 2.4118298031989084e-05, + "loss": 0.0163, + "step": 3930 + }, + { + "epoch": 6.599664991624791, + "grad_norm": 0.07965271174907684, + "learning_rate": 2.4085441085835534e-05, + "loss": 0.0162, + "step": 3940 + }, + { + "epoch": 6.61641541038526, + "grad_norm": 0.10382688790559769, + "learning_rate": 2.405251514179604e-05, + "loss": 0.0172, + "step": 3950 + }, + { + "epoch": 6.633165829145729, + "grad_norm": 0.0887349396944046, + "learning_rate": 2.401952044992126e-05, + "loss": 0.0181, + "step": 3960 + }, + { + "epoch": 6.649916247906198, + "grad_norm": 0.10867097228765488, + "learning_rate": 2.3986457260783942e-05, + "loss": 0.016, + "step": 3970 + }, + { + "epoch": 6.666666666666667, + "grad_norm": 0.10828111320734024, + "learning_rate": 2.395332582547703e-05, + "loss": 0.0216, + "step": 3980 + }, + { + "epoch": 6.683417085427136, + "grad_norm": 0.08616909384727478, + "learning_rate": 2.3920126395611755e-05, + "loss": 0.019, + "step": 3990 + }, + { + "epoch": 6.700167504187605, + "grad_norm": 0.10397583991289139, + "learning_rate": 2.3886859223315725e-05, + "loss": 0.0199, + "step": 4000 + }, + { + "epoch": 6.716917922948074, + "grad_norm": 0.12109193205833435, + "learning_rate": 2.385352456123099e-05, + "loss": 0.0191, + "step": 4010 + }, + { + "epoch": 6.733668341708543, + "grad_norm": 0.07098208367824554, + "learning_rate": 2.3820122662512155e-05, + "loss": 0.0164, + "step": 4020 + }, + { + "epoch": 6.750418760469012, + "grad_norm": 0.1095917746424675, + "learning_rate": 2.3786653780824443e-05, + "loss": 0.0153, + "step": 4030 + }, + { + "epoch": 6.767169179229481, + "grad_norm": 0.08910483866930008, + "learning_rate": 2.3753118170341762e-05, + "loss": 0.0168, + "step": 4040 + }, + { + "epoch": 6.78391959798995, + "grad_norm": 0.07048118114471436, + "learning_rate": 2.3719516085744782e-05, + "loss": 0.0152, + "step": 4050 + }, + { + "epoch": 6.800670016750419, + "grad_norm": 0.1071673035621643, + "learning_rate": 2.3685847782218997e-05, + "loss": 0.0158, + "step": 4060 + }, + { + "epoch": 6.817420435510888, + "grad_norm": 0.10280796140432358, + "learning_rate": 2.3652113515452792e-05, + "loss": 0.0173, + "step": 4070 + }, + { + "epoch": 6.834170854271357, + "grad_norm": 0.10277356952428818, + "learning_rate": 2.3618313541635506e-05, + "loss": 0.0173, + "step": 4080 + }, + { + "epoch": 6.850921273031826, + "grad_norm": 0.09373802691698074, + "learning_rate": 2.358444811745547e-05, + "loss": 0.0166, + "step": 4090 + }, + { + "epoch": 6.867671691792295, + "grad_norm": 0.09374375641345978, + "learning_rate": 2.3550517500098068e-05, + "loss": 0.0171, + "step": 4100 + }, + { + "epoch": 6.884422110552764, + "grad_norm": 0.08589532971382141, + "learning_rate": 2.3516521947243787e-05, + "loss": 0.0177, + "step": 4110 + }, + { + "epoch": 6.901172529313233, + "grad_norm": 0.09759582579135895, + "learning_rate": 2.348246171706626e-05, + "loss": 0.0161, + "step": 4120 + }, + { + "epoch": 6.917922948073702, + "grad_norm": 0.08552340418100357, + "learning_rate": 2.3448337068230288e-05, + "loss": 0.0156, + "step": 4130 + }, + { + "epoch": 6.934673366834171, + "grad_norm": 0.07144278287887573, + "learning_rate": 2.3414148259889896e-05, + "loss": 0.0184, + "step": 4140 + }, + { + "epoch": 6.95142378559464, + "grad_norm": 0.08075377345085144, + "learning_rate": 2.3379895551686358e-05, + "loss": 0.0182, + "step": 4150 + }, + { + "epoch": 6.968174204355109, + "grad_norm": 0.09423919767141342, + "learning_rate": 2.334557920374623e-05, + "loss": 0.0173, + "step": 4160 + }, + { + "epoch": 6.984924623115578, + "grad_norm": 0.09013880789279938, + "learning_rate": 2.3311199476679367e-05, + "loss": 0.0172, + "step": 4170 + }, + { + "epoch": 7.001675041876047, + "grad_norm": 0.1004585400223732, + "learning_rate": 2.327675663157693e-05, + "loss": 0.0171, + "step": 4180 + }, + { + "epoch": 7.018425460636516, + "grad_norm": 0.09630786627531052, + "learning_rate": 2.3242250930009446e-05, + "loss": 0.0169, + "step": 4190 + }, + { + "epoch": 7.035175879396985, + "grad_norm": 0.09441886842250824, + "learning_rate": 2.3207682634024777e-05, + "loss": 0.016, + "step": 4200 + }, + { + "epoch": 7.051926298157454, + "grad_norm": 0.1110304594039917, + "learning_rate": 2.3173052006146155e-05, + "loss": 0.0163, + "step": 4210 + }, + { + "epoch": 7.068676716917923, + "grad_norm": 0.07521197199821472, + "learning_rate": 2.313835930937018e-05, + "loss": 0.0146, + "step": 4220 + }, + { + "epoch": 7.085427135678392, + "grad_norm": 0.06828752905130386, + "learning_rate": 2.310360480716483e-05, + "loss": 0.0156, + "step": 4230 + }, + { + "epoch": 7.102177554438861, + "grad_norm": 0.08863577991724014, + "learning_rate": 2.3068788763467434e-05, + "loss": 0.0154, + "step": 4240 + }, + { + "epoch": 7.11892797319933, + "grad_norm": 0.10143563151359558, + "learning_rate": 2.3033911442682708e-05, + "loss": 0.0171, + "step": 4250 + }, + { + "epoch": 7.135678391959799, + "grad_norm": 0.10039176791906357, + "learning_rate": 2.299897310968072e-05, + "loss": 0.0185, + "step": 4260 + }, + { + "epoch": 7.152428810720268, + "grad_norm": 0.06830760836601257, + "learning_rate": 2.2963974029794893e-05, + "loss": 0.0135, + "step": 4270 + }, + { + "epoch": 7.169179229480737, + "grad_norm": 0.06696446985006332, + "learning_rate": 2.2928914468819968e-05, + "loss": 0.016, + "step": 4280 + }, + { + "epoch": 7.185929648241206, + "grad_norm": 0.06527981907129288, + "learning_rate": 2.2893794693010002e-05, + "loss": 0.0149, + "step": 4290 + }, + { + "epoch": 7.202680067001675, + "grad_norm": 0.08876840025186539, + "learning_rate": 2.2858614969076358e-05, + "loss": 0.0179, + "step": 4300 + }, + { + "epoch": 7.219430485762144, + "grad_norm": 0.08040497452020645, + "learning_rate": 2.2823375564185645e-05, + "loss": 0.0172, + "step": 4310 + }, + { + "epoch": 7.236180904522613, + "grad_norm": 0.06119128316640854, + "learning_rate": 2.278807674595773e-05, + "loss": 0.0157, + "step": 4320 + }, + { + "epoch": 7.252931323283082, + "grad_norm": 0.07552856206893921, + "learning_rate": 2.275271878246367e-05, + "loss": 0.0161, + "step": 4330 + }, + { + "epoch": 7.269681742043551, + "grad_norm": 0.09174931794404984, + "learning_rate": 2.271730194222369e-05, + "loss": 0.0157, + "step": 4340 + }, + { + "epoch": 7.28643216080402, + "grad_norm": 0.10719872266054153, + "learning_rate": 2.2681826494205156e-05, + "loss": 0.0175, + "step": 4350 + }, + { + "epoch": 7.303182579564489, + "grad_norm": 0.08944938331842422, + "learning_rate": 2.2646292707820513e-05, + "loss": 0.0146, + "step": 4360 + }, + { + "epoch": 7.319932998324958, + "grad_norm": 0.08042101562023163, + "learning_rate": 2.2610700852925254e-05, + "loss": 0.0145, + "step": 4370 + }, + { + "epoch": 7.336683417085427, + "grad_norm": 0.10604419559240341, + "learning_rate": 2.2575051199815854e-05, + "loss": 0.0175, + "step": 4380 + }, + { + "epoch": 7.353433835845896, + "grad_norm": 0.07255196571350098, + "learning_rate": 2.253934401922773e-05, + "loss": 0.0161, + "step": 4390 + }, + { + "epoch": 7.370184254606365, + "grad_norm": 0.07271875441074371, + "learning_rate": 2.250357958233319e-05, + "loss": 0.0181, + "step": 4400 + }, + { + "epoch": 7.386934673366834, + "grad_norm": 0.08586633205413818, + "learning_rate": 2.2467758160739352e-05, + "loss": 0.0159, + "step": 4410 + }, + { + "epoch": 7.403685092127303, + "grad_norm": 0.07192263007164001, + "learning_rate": 2.2431880026486114e-05, + "loss": 0.0176, + "step": 4420 + }, + { + "epoch": 7.420435510887772, + "grad_norm": 0.0814385861158371, + "learning_rate": 2.2395945452044045e-05, + "loss": 0.0171, + "step": 4430 + }, + { + "epoch": 7.437185929648241, + "grad_norm": 0.0670303925871849, + "learning_rate": 2.2359954710312363e-05, + "loss": 0.0145, + "step": 4440 + }, + { + "epoch": 7.45393634840871, + "grad_norm": 0.08451966196298599, + "learning_rate": 2.232390807461682e-05, + "loss": 0.0148, + "step": 4450 + }, + { + "epoch": 7.4706867671691795, + "grad_norm": 0.09510994702577591, + "learning_rate": 2.2287805818707658e-05, + "loss": 0.016, + "step": 4460 + }, + { + "epoch": 7.4874371859296485, + "grad_norm": 0.08456354588270187, + "learning_rate": 2.2251648216757514e-05, + "loss": 0.014, + "step": 4470 + }, + { + "epoch": 7.5041876046901175, + "grad_norm": 0.11731875687837601, + "learning_rate": 2.2215435543359337e-05, + "loss": 0.0154, + "step": 4480 + }, + { + "epoch": 7.5209380234505865, + "grad_norm": 0.10068397969007492, + "learning_rate": 2.2179168073524308e-05, + "loss": 0.0161, + "step": 4490 + }, + { + "epoch": 7.5376884422110555, + "grad_norm": 0.09683054685592651, + "learning_rate": 2.2142846082679763e-05, + "loss": 0.0168, + "step": 4500 + }, + { + "epoch": 7.5544388609715245, + "grad_norm": 0.06540673226118088, + "learning_rate": 2.2106469846667078e-05, + "loss": 0.013, + "step": 4510 + }, + { + "epoch": 7.5711892797319935, + "grad_norm": 0.0816856399178505, + "learning_rate": 2.2070039641739578e-05, + "loss": 0.0154, + "step": 4520 + }, + { + "epoch": 7.5879396984924625, + "grad_norm": 0.07817541062831879, + "learning_rate": 2.2033555744560462e-05, + "loss": 0.017, + "step": 4530 + }, + { + "epoch": 7.6046901172529315, + "grad_norm": 0.08995489031076431, + "learning_rate": 2.1997018432200674e-05, + "loss": 0.0153, + "step": 4540 + }, + { + "epoch": 7.6214405360134005, + "grad_norm": 0.09150712937116623, + "learning_rate": 2.1960427982136818e-05, + "loss": 0.0154, + "step": 4550 + }, + { + "epoch": 7.63819095477387, + "grad_norm": 0.11141885071992874, + "learning_rate": 2.1923784672249038e-05, + "loss": 0.0159, + "step": 4560 + }, + { + "epoch": 7.654941373534339, + "grad_norm": 0.09261714667081833, + "learning_rate": 2.1887088780818918e-05, + "loss": 0.015, + "step": 4570 + }, + { + "epoch": 7.671691792294808, + "grad_norm": 0.11019844561815262, + "learning_rate": 2.1850340586527354e-05, + "loss": 0.0181, + "step": 4580 + }, + { + "epoch": 7.688442211055277, + "grad_norm": 0.09286614507436752, + "learning_rate": 2.1813540368452468e-05, + "loss": 0.0165, + "step": 4590 + }, + { + "epoch": 7.705192629815746, + "grad_norm": 0.07964478433132172, + "learning_rate": 2.1776688406067447e-05, + "loss": 0.0162, + "step": 4600 + }, + { + "epoch": 7.721943048576215, + "grad_norm": 0.09548459202051163, + "learning_rate": 2.1739784979238447e-05, + "loss": 0.0165, + "step": 4610 + }, + { + "epoch": 7.738693467336684, + "grad_norm": 0.06440986692905426, + "learning_rate": 2.1702830368222478e-05, + "loss": 0.0141, + "step": 4620 + }, + { + "epoch": 7.755443886097153, + "grad_norm": 0.07117027044296265, + "learning_rate": 2.166582485366523e-05, + "loss": 0.0147, + "step": 4630 + }, + { + "epoch": 7.772194304857622, + "grad_norm": 0.08931209146976471, + "learning_rate": 2.1628768716599e-05, + "loss": 0.0147, + "step": 4640 + }, + { + "epoch": 7.788944723618091, + "grad_norm": 0.09026764333248138, + "learning_rate": 2.1591662238440515e-05, + "loss": 0.0159, + "step": 4650 + }, + { + "epoch": 7.80569514237856, + "grad_norm": 0.1498933583498001, + "learning_rate": 2.1554505700988813e-05, + "loss": 0.0155, + "step": 4660 + }, + { + "epoch": 7.822445561139029, + "grad_norm": 0.06725804507732391, + "learning_rate": 2.1517299386423094e-05, + "loss": 0.016, + "step": 4670 + }, + { + "epoch": 7.839195979899498, + "grad_norm": 0.08815880864858627, + "learning_rate": 2.148004357730059e-05, + "loss": 0.0151, + "step": 4680 + }, + { + "epoch": 7.855946398659967, + "grad_norm": 0.06878040730953217, + "learning_rate": 2.1442738556554404e-05, + "loss": 0.0172, + "step": 4690 + }, + { + "epoch": 7.872696817420436, + "grad_norm": 0.06804270297288895, + "learning_rate": 2.1405384607491375e-05, + "loss": 0.0156, + "step": 4700 + }, + { + "epoch": 7.889447236180905, + "grad_norm": 0.070171058177948, + "learning_rate": 2.1367982013789913e-05, + "loss": 0.0153, + "step": 4710 + }, + { + "epoch": 7.906197654941374, + "grad_norm": 0.07167346775531769, + "learning_rate": 2.133053105949786e-05, + "loss": 0.0156, + "step": 4720 + }, + { + "epoch": 7.922948073701843, + "grad_norm": 0.09849503636360168, + "learning_rate": 2.129303202903032e-05, + "loss": 0.0165, + "step": 4730 + }, + { + "epoch": 7.939698492462312, + "grad_norm": 0.09628047794103622, + "learning_rate": 2.1255485207167504e-05, + "loss": 0.016, + "step": 4740 + }, + { + "epoch": 7.956448911222781, + "grad_norm": 0.07518996298313141, + "learning_rate": 2.1217890879052556e-05, + "loss": 0.0148, + "step": 4750 + }, + { + "epoch": 7.97319932998325, + "grad_norm": 0.09254742413759232, + "learning_rate": 2.1180249330189423e-05, + "loss": 0.0177, + "step": 4760 + }, + { + "epoch": 7.989949748743719, + "grad_norm": 0.10075143724679947, + "learning_rate": 2.1142560846440645e-05, + "loss": 0.0173, + "step": 4770 + }, + { + "epoch": 8.006700167504187, + "grad_norm": 0.0872223973274231, + "learning_rate": 2.1104825714025203e-05, + "loss": 0.0155, + "step": 4780 + }, + { + "epoch": 8.023450586264657, + "grad_norm": 0.07023284584283829, + "learning_rate": 2.106704421951634e-05, + "loss": 0.0169, + "step": 4790 + }, + { + "epoch": 8.040201005025125, + "grad_norm": 0.0736195370554924, + "learning_rate": 2.1029216649839404e-05, + "loss": 0.0153, + "step": 4800 + }, + { + "epoch": 8.056951423785595, + "grad_norm": 0.12984727323055267, + "learning_rate": 2.099134329226963e-05, + "loss": 0.0194, + "step": 4810 + }, + { + "epoch": 8.073701842546063, + "grad_norm": 0.06633501499891281, + "learning_rate": 2.0953424434430006e-05, + "loss": 0.0172, + "step": 4820 + }, + { + "epoch": 8.090452261306533, + "grad_norm": 0.07725346088409424, + "learning_rate": 2.091546036428905e-05, + "loss": 0.0167, + "step": 4830 + }, + { + "epoch": 8.107202680067001, + "grad_norm": 0.0776301696896553, + "learning_rate": 2.0877451370158633e-05, + "loss": 0.0167, + "step": 4840 + }, + { + "epoch": 8.12395309882747, + "grad_norm": 0.08715241402387619, + "learning_rate": 2.083939774069181e-05, + "loss": 0.0152, + "step": 4850 + }, + { + "epoch": 8.140703517587939, + "grad_norm": 0.05368698760867119, + "learning_rate": 2.080129976488059e-05, + "loss": 0.0158, + "step": 4860 + }, + { + "epoch": 8.157453936348409, + "grad_norm": 0.07227354496717453, + "learning_rate": 2.0763157732053795e-05, + "loss": 0.0167, + "step": 4870 + }, + { + "epoch": 8.174204355108877, + "grad_norm": 0.08462859690189362, + "learning_rate": 2.0724971931874796e-05, + "loss": 0.0155, + "step": 4880 + }, + { + "epoch": 8.190954773869347, + "grad_norm": 0.0951157882809639, + "learning_rate": 2.0686742654339363e-05, + "loss": 0.0151, + "step": 4890 + }, + { + "epoch": 8.207705192629815, + "grad_norm": 0.07463134825229645, + "learning_rate": 2.0648470189773455e-05, + "loss": 0.0136, + "step": 4900 + }, + { + "epoch": 8.224455611390285, + "grad_norm": 0.08849921822547913, + "learning_rate": 2.0610154828830997e-05, + "loss": 0.0145, + "step": 4910 + }, + { + "epoch": 8.241206030150753, + "grad_norm": 0.08144073933362961, + "learning_rate": 2.0571796862491686e-05, + "loss": 0.0173, + "step": 4920 + }, + { + "epoch": 8.257956448911223, + "grad_norm": 0.08935582637786865, + "learning_rate": 2.0533396582058783e-05, + "loss": 0.0173, + "step": 4930 + }, + { + "epoch": 8.274706867671691, + "grad_norm": 0.06497425585985184, + "learning_rate": 2.0494954279156893e-05, + "loss": 0.0159, + "step": 4940 + }, + { + "epoch": 8.291457286432161, + "grad_norm": 0.09040762484073639, + "learning_rate": 2.0456470245729748e-05, + "loss": 0.018, + "step": 4950 + }, + { + "epoch": 8.308207705192629, + "grad_norm": 0.11972896009683609, + "learning_rate": 2.0417944774038013e-05, + "loss": 0.016, + "step": 4960 + }, + { + "epoch": 8.324958123953099, + "grad_norm": 0.09560854732990265, + "learning_rate": 2.0379378156657026e-05, + "loss": 0.0151, + "step": 4970 + }, + { + "epoch": 8.341708542713567, + "grad_norm": 0.09106440097093582, + "learning_rate": 2.0340770686474618e-05, + "loss": 0.0169, + "step": 4980 + }, + { + "epoch": 8.358458961474037, + "grad_norm": 0.07660805433988571, + "learning_rate": 2.030212265668886e-05, + "loss": 0.0141, + "step": 4990 + }, + { + "epoch": 8.375209380234505, + "grad_norm": 0.08468059450387955, + "learning_rate": 2.0263434360805847e-05, + "loss": 0.0167, + "step": 5000 + }, + { + "epoch": 8.391959798994975, + "grad_norm": 0.0799565464258194, + "learning_rate": 2.0224706092637472e-05, + "loss": 0.0153, + "step": 5010 + }, + { + "epoch": 8.408710217755443, + "grad_norm": 0.10247821360826492, + "learning_rate": 2.018593814629919e-05, + "loss": 0.0168, + "step": 5020 + }, + { + "epoch": 8.425460636515913, + "grad_norm": 0.07618727535009384, + "learning_rate": 2.0147130816207786e-05, + "loss": 0.016, + "step": 5030 + }, + { + "epoch": 8.442211055276381, + "grad_norm": 0.11073528230190277, + "learning_rate": 2.0108284397079127e-05, + "loss": 0.015, + "step": 5040 + }, + { + "epoch": 8.458961474036851, + "grad_norm": 0.10041352361440659, + "learning_rate": 2.006939918392596e-05, + "loss": 0.0156, + "step": 5050 + }, + { + "epoch": 8.47571189279732, + "grad_norm": 0.06778070330619812, + "learning_rate": 2.0030475472055625e-05, + "loss": 0.0151, + "step": 5060 + }, + { + "epoch": 8.492462311557789, + "grad_norm": 0.07777587324380875, + "learning_rate": 1.9991513557067835e-05, + "loss": 0.0144, + "step": 5070 + }, + { + "epoch": 8.509212730318257, + "grad_norm": 0.06961089372634888, + "learning_rate": 1.9952513734852444e-05, + "loss": 0.0139, + "step": 5080 + }, + { + "epoch": 8.525963149078727, + "grad_norm": 0.07862704992294312, + "learning_rate": 1.991347630158717e-05, + "loss": 0.0148, + "step": 5090 + }, + { + "epoch": 8.542713567839195, + "grad_norm": 0.10534326732158661, + "learning_rate": 1.9874401553735385e-05, + "loss": 0.0161, + "step": 5100 + }, + { + "epoch": 8.559463986599665, + "grad_norm": 0.08768954873085022, + "learning_rate": 1.9835289788043814e-05, + "loss": 0.0137, + "step": 5110 + }, + { + "epoch": 8.576214405360133, + "grad_norm": 0.06274593621492386, + "learning_rate": 1.979614130154032e-05, + "loss": 0.0149, + "step": 5120 + }, + { + "epoch": 8.592964824120603, + "grad_norm": 0.0664563998579979, + "learning_rate": 1.9756956391531646e-05, + "loss": 0.013, + "step": 5130 + }, + { + "epoch": 8.609715242881071, + "grad_norm": 0.061676476150751114, + "learning_rate": 1.9717735355601126e-05, + "loss": 0.0162, + "step": 5140 + }, + { + "epoch": 8.626465661641541, + "grad_norm": 0.08172836154699326, + "learning_rate": 1.9678478491606462e-05, + "loss": 0.0174, + "step": 5150 + }, + { + "epoch": 8.64321608040201, + "grad_norm": 0.05784829333424568, + "learning_rate": 1.9639186097677442e-05, + "loss": 0.0149, + "step": 5160 + }, + { + "epoch": 8.65996649916248, + "grad_norm": 0.08542623370885849, + "learning_rate": 1.9599858472213682e-05, + "loss": 0.0147, + "step": 5170 + }, + { + "epoch": 8.676716917922947, + "grad_norm": 0.07957877963781357, + "learning_rate": 1.9560495913882353e-05, + "loss": 0.0133, + "step": 5180 + }, + { + "epoch": 8.693467336683417, + "grad_norm": 0.0810636356472969, + "learning_rate": 1.9521098721615923e-05, + "loss": 0.0145, + "step": 5190 + }, + { + "epoch": 8.710217755443885, + "grad_norm": 0.07732214033603668, + "learning_rate": 1.9481667194609882e-05, + "loss": 0.0145, + "step": 5200 + }, + { + "epoch": 8.726968174204355, + "grad_norm": 0.0780918151140213, + "learning_rate": 1.9442201632320464e-05, + "loss": 0.0147, + "step": 5210 + }, + { + "epoch": 8.743718592964823, + "grad_norm": 0.09253431856632233, + "learning_rate": 1.9402702334462383e-05, + "loss": 0.0155, + "step": 5220 + }, + { + "epoch": 8.760469011725293, + "grad_norm": 0.06418832391500473, + "learning_rate": 1.9363169601006552e-05, + "loss": 0.0126, + "step": 5230 + }, + { + "epoch": 8.777219430485761, + "grad_norm": 0.09778792411088943, + "learning_rate": 1.9323603732177798e-05, + "loss": 0.0146, + "step": 5240 + }, + { + "epoch": 8.793969849246231, + "grad_norm": 0.0891941487789154, + "learning_rate": 1.9284005028452607e-05, + "loss": 0.0141, + "step": 5250 + }, + { + "epoch": 8.8107202680067, + "grad_norm": 0.1019863709807396, + "learning_rate": 1.9244373790556803e-05, + "loss": 0.0154, + "step": 5260 + }, + { + "epoch": 8.82747068676717, + "grad_norm": 0.08988063782453537, + "learning_rate": 1.9204710319463297e-05, + "loss": 0.0151, + "step": 5270 + }, + { + "epoch": 8.844221105527637, + "grad_norm": 0.0668616071343422, + "learning_rate": 1.9165014916389792e-05, + "loss": 0.0128, + "step": 5280 + }, + { + "epoch": 8.860971524288107, + "grad_norm": 0.08318133652210236, + "learning_rate": 1.9125287882796483e-05, + "loss": 0.0159, + "step": 5290 + }, + { + "epoch": 8.877721943048575, + "grad_norm": 0.07382380962371826, + "learning_rate": 1.9085529520383783e-05, + "loss": 0.0142, + "step": 5300 + }, + { + "epoch": 8.894472361809045, + "grad_norm": 0.07555682957172394, + "learning_rate": 1.9045740131090037e-05, + "loss": 0.0142, + "step": 5310 + }, + { + "epoch": 8.911222780569513, + "grad_norm": 0.09456376731395721, + "learning_rate": 1.9005920017089196e-05, + "loss": 0.0139, + "step": 5320 + }, + { + "epoch": 8.927973199329983, + "grad_norm": 0.11305142194032669, + "learning_rate": 1.8966069480788562e-05, + "loss": 0.0146, + "step": 5330 + }, + { + "epoch": 8.944723618090451, + "grad_norm": 0.06259690970182419, + "learning_rate": 1.8926188824826477e-05, + "loss": 0.0154, + "step": 5340 + }, + { + "epoch": 8.961474036850921, + "grad_norm": 0.10811670124530792, + "learning_rate": 1.8886278352070002e-05, + "loss": 0.014, + "step": 5350 + }, + { + "epoch": 8.97822445561139, + "grad_norm": 0.06475009024143219, + "learning_rate": 1.884633836561265e-05, + "loss": 0.014, + "step": 5360 + }, + { + "epoch": 8.99497487437186, + "grad_norm": 0.08297079056501389, + "learning_rate": 1.8806369168772086e-05, + "loss": 0.0141, + "step": 5370 + }, + { + "epoch": 9.011725293132328, + "grad_norm": 0.079750195145607, + "learning_rate": 1.8766371065087778e-05, + "loss": 0.0142, + "step": 5380 + }, + { + "epoch": 9.028475711892797, + "grad_norm": 0.07751394808292389, + "learning_rate": 1.8726344358318736e-05, + "loss": 0.0155, + "step": 5390 + }, + { + "epoch": 9.045226130653266, + "grad_norm": 0.0738295242190361, + "learning_rate": 1.86862893524412e-05, + "loss": 0.0142, + "step": 5400 + }, + { + "epoch": 9.061976549413735, + "grad_norm": 0.11355958878993988, + "learning_rate": 1.864620635164631e-05, + "loss": 0.0161, + "step": 5410 + }, + { + "epoch": 9.078726968174204, + "grad_norm": 0.08234552294015884, + "learning_rate": 1.8606095660337813e-05, + "loss": 0.016, + "step": 5420 + }, + { + "epoch": 9.095477386934673, + "grad_norm": 0.09014493972063065, + "learning_rate": 1.8565957583129753e-05, + "loss": 0.0133, + "step": 5430 + }, + { + "epoch": 9.112227805695142, + "grad_norm": 0.08228293806314468, + "learning_rate": 1.852579242484414e-05, + "loss": 0.0152, + "step": 5440 + }, + { + "epoch": 9.128978224455611, + "grad_norm": 0.06658344715833664, + "learning_rate": 1.848560049050866e-05, + "loss": 0.0158, + "step": 5450 + }, + { + "epoch": 9.14572864321608, + "grad_norm": 0.08792341500520706, + "learning_rate": 1.8445382085354336e-05, + "loss": 0.0122, + "step": 5460 + }, + { + "epoch": 9.16247906197655, + "grad_norm": 0.07732446491718292, + "learning_rate": 1.8405137514813224e-05, + "loss": 0.0141, + "step": 5470 + }, + { + "epoch": 9.179229480737018, + "grad_norm": 0.06328639388084412, + "learning_rate": 1.8364867084516084e-05, + "loss": 0.0138, + "step": 5480 + }, + { + "epoch": 9.195979899497488, + "grad_norm": 0.05957376956939697, + "learning_rate": 1.8324571100290064e-05, + "loss": 0.0137, + "step": 5490 + }, + { + "epoch": 9.212730318257956, + "grad_norm": 0.0765262320637703, + "learning_rate": 1.8284249868156378e-05, + "loss": 0.0153, + "step": 5500 + }, + { + "epoch": 9.229480737018426, + "grad_norm": 0.0667581558227539, + "learning_rate": 1.8243903694327992e-05, + "loss": 0.0139, + "step": 5510 + }, + { + "epoch": 9.246231155778894, + "grad_norm": 0.07833325862884521, + "learning_rate": 1.8203532885207272e-05, + "loss": 0.0156, + "step": 5520 + }, + { + "epoch": 9.262981574539364, + "grad_norm": 0.06736069172620773, + "learning_rate": 1.8163137747383675e-05, + "loss": 0.0127, + "step": 5530 + }, + { + "epoch": 9.279731993299832, + "grad_norm": 0.0944928526878357, + "learning_rate": 1.812271858763143e-05, + "loss": 0.0144, + "step": 5540 + }, + { + "epoch": 9.296482412060302, + "grad_norm": 0.09697055071592331, + "learning_rate": 1.8082275712907187e-05, + "loss": 0.0151, + "step": 5550 + }, + { + "epoch": 9.31323283082077, + "grad_norm": 0.09629473835229874, + "learning_rate": 1.8041809430347688e-05, + "loss": 0.0176, + "step": 5560 + }, + { + "epoch": 9.32998324958124, + "grad_norm": 0.07615918666124344, + "learning_rate": 1.8001320047267465e-05, + "loss": 0.0148, + "step": 5570 + }, + { + "epoch": 9.346733668341708, + "grad_norm": 0.07980474084615707, + "learning_rate": 1.796080787115646e-05, + "loss": 0.0149, + "step": 5580 + }, + { + "epoch": 9.363484087102178, + "grad_norm": 0.10057661682367325, + "learning_rate": 1.7920273209677722e-05, + "loss": 0.0167, + "step": 5590 + }, + { + "epoch": 9.380234505862646, + "grad_norm": 0.0847344771027565, + "learning_rate": 1.7879716370665068e-05, + "loss": 0.016, + "step": 5600 + }, + { + "epoch": 9.396984924623116, + "grad_norm": 0.07230829447507858, + "learning_rate": 1.783913766212073e-05, + "loss": 0.0161, + "step": 5610 + }, + { + "epoch": 9.413735343383584, + "grad_norm": 0.0874628946185112, + "learning_rate": 1.779853739221302e-05, + "loss": 0.0161, + "step": 5620 + }, + { + "epoch": 9.430485762144054, + "grad_norm": 0.07002632319927216, + "learning_rate": 1.775791586927401e-05, + "loss": 0.0148, + "step": 5630 + }, + { + "epoch": 9.447236180904522, + "grad_norm": 0.09408432990312576, + "learning_rate": 1.771727340179716e-05, + "loss": 0.013, + "step": 5640 + }, + { + "epoch": 9.463986599664992, + "grad_norm": 0.07826568186283112, + "learning_rate": 1.7676610298434993e-05, + "loss": 0.0155, + "step": 5650 + }, + { + "epoch": 9.48073701842546, + "grad_norm": 0.09074067324399948, + "learning_rate": 1.7635926867996758e-05, + "loss": 0.0144, + "step": 5660 + }, + { + "epoch": 9.49748743718593, + "grad_norm": 0.10472352802753448, + "learning_rate": 1.7595223419446064e-05, + "loss": 0.0159, + "step": 5670 + }, + { + "epoch": 9.514237855946398, + "grad_norm": 0.05583181977272034, + "learning_rate": 1.7554500261898545e-05, + "loss": 0.0138, + "step": 5680 + }, + { + "epoch": 9.530988274706868, + "grad_norm": 0.07133681327104568, + "learning_rate": 1.7513757704619513e-05, + "loss": 0.016, + "step": 5690 + }, + { + "epoch": 9.547738693467338, + "grad_norm": 0.057500433176755905, + "learning_rate": 1.747299605702161e-05, + "loss": 0.0135, + "step": 5700 + }, + { + "epoch": 9.564489112227806, + "grad_norm": 0.10911339521408081, + "learning_rate": 1.7432215628662463e-05, + "loss": 0.014, + "step": 5710 + }, + { + "epoch": 9.581239530988274, + "grad_norm": 0.09471974521875381, + "learning_rate": 1.7391416729242307e-05, + "loss": 0.0142, + "step": 5720 + }, + { + "epoch": 9.597989949748744, + "grad_norm": 0.08386453986167908, + "learning_rate": 1.7350599668601666e-05, + "loss": 0.0152, + "step": 5730 + }, + { + "epoch": 9.614740368509214, + "grad_norm": 0.09431836754083633, + "learning_rate": 1.730976475671899e-05, + "loss": 0.0181, + "step": 5740 + }, + { + "epoch": 9.631490787269682, + "grad_norm": 0.07204046100378036, + "learning_rate": 1.726891230370829e-05, + "loss": 0.0138, + "step": 5750 + }, + { + "epoch": 9.64824120603015, + "grad_norm": 0.08191115409135818, + "learning_rate": 1.7228042619816786e-05, + "loss": 0.0132, + "step": 5760 + }, + { + "epoch": 9.66499162479062, + "grad_norm": 0.06609786301851273, + "learning_rate": 1.7187156015422573e-05, + "loss": 0.0146, + "step": 5770 + }, + { + "epoch": 9.68174204355109, + "grad_norm": 0.08089634031057358, + "learning_rate": 1.7146252801032225e-05, + "loss": 0.014, + "step": 5780 + }, + { + "epoch": 9.698492462311558, + "grad_norm": 0.06636945903301239, + "learning_rate": 1.7105333287278465e-05, + "loss": 0.0126, + "step": 5790 + }, + { + "epoch": 9.715242881072026, + "grad_norm": 0.08253764361143112, + "learning_rate": 1.706439778491781e-05, + "loss": 0.0146, + "step": 5800 + }, + { + "epoch": 9.731993299832496, + "grad_norm": 0.06800912320613861, + "learning_rate": 1.702344660482818e-05, + "loss": 0.015, + "step": 5810 + }, + { + "epoch": 9.748743718592966, + "grad_norm": 0.05859517678618431, + "learning_rate": 1.6982480058006575e-05, + "loss": 0.0166, + "step": 5820 + }, + { + "epoch": 9.765494137353434, + "grad_norm": 0.09688255935907364, + "learning_rate": 1.6941498455566684e-05, + "loss": 0.0134, + "step": 5830 + }, + { + "epoch": 9.782244556113902, + "grad_norm": 0.07653312385082245, + "learning_rate": 1.690050210873654e-05, + "loss": 0.014, + "step": 5840 + }, + { + "epoch": 9.798994974874372, + "grad_norm": 0.08134609460830688, + "learning_rate": 1.6859491328856143e-05, + "loss": 0.0133, + "step": 5850 + }, + { + "epoch": 9.815745393634842, + "grad_norm": 0.06555905193090439, + "learning_rate": 1.6818466427375114e-05, + "loss": 0.0149, + "step": 5860 + }, + { + "epoch": 9.83249581239531, + "grad_norm": 0.06328080594539642, + "learning_rate": 1.67774277158503e-05, + "loss": 0.0134, + "step": 5870 + }, + { + "epoch": 9.849246231155778, + "grad_norm": 0.08666656166315079, + "learning_rate": 1.673637550594344e-05, + "loss": 0.0146, + "step": 5880 + }, + { + "epoch": 9.865996649916248, + "grad_norm": 0.06649037450551987, + "learning_rate": 1.669531010941879e-05, + "loss": 0.013, + "step": 5890 + }, + { + "epoch": 9.882747068676718, + "grad_norm": 0.0830390453338623, + "learning_rate": 1.6654231838140723e-05, + "loss": 0.013, + "step": 5900 + }, + { + "epoch": 9.899497487437186, + "grad_norm": 0.07208152860403061, + "learning_rate": 1.661314100407142e-05, + "loss": 0.0141, + "step": 5910 + }, + { + "epoch": 9.916247906197654, + "grad_norm": 0.07338222116231918, + "learning_rate": 1.657203791926845e-05, + "loss": 0.0123, + "step": 5920 + }, + { + "epoch": 9.932998324958124, + "grad_norm": 0.09653204679489136, + "learning_rate": 1.653092289588241e-05, + "loss": 0.0144, + "step": 5930 + }, + { + "epoch": 9.949748743718594, + "grad_norm": 0.07927566021680832, + "learning_rate": 1.6489796246154586e-05, + "loss": 0.0143, + "step": 5940 + }, + { + "epoch": 9.966499162479062, + "grad_norm": 0.057749975472688675, + "learning_rate": 1.6448658282414543e-05, + "loss": 0.0139, + "step": 5950 + }, + { + "epoch": 9.98324958123953, + "grad_norm": 0.07733950763940811, + "learning_rate": 1.6407509317077764e-05, + "loss": 0.0145, + "step": 5960 + }, + { + "epoch": 10.0, + "grad_norm": 0.06586837023496628, + "learning_rate": 1.6366349662643297e-05, + "loss": 0.0135, + "step": 5970 + }, + { + "epoch": 10.01675041876047, + "grad_norm": 0.09854725748300552, + "learning_rate": 1.6325179631691355e-05, + "loss": 0.0145, + "step": 5980 + }, + { + "epoch": 10.033500837520938, + "grad_norm": 0.08377289772033691, + "learning_rate": 1.6283999536880957e-05, + "loss": 0.0136, + "step": 5990 + }, + { + "epoch": 10.050251256281408, + "grad_norm": 0.07842150330543518, + "learning_rate": 1.6242809690947555e-05, + "loss": 0.0123, + "step": 6000 + }, + { + "epoch": 10.067001675041876, + "grad_norm": 0.08640950173139572, + "learning_rate": 1.6201610406700645e-05, + "loss": 0.0133, + "step": 6010 + }, + { + "epoch": 10.083752093802346, + "grad_norm": 0.1089024543762207, + "learning_rate": 1.6160401997021406e-05, + "loss": 0.014, + "step": 6020 + }, + { + "epoch": 10.100502512562814, + "grad_norm": 0.06064023822546005, + "learning_rate": 1.6119184774860322e-05, + "loss": 0.0146, + "step": 6030 + }, + { + "epoch": 10.117252931323284, + "grad_norm": 0.06450966000556946, + "learning_rate": 1.6077959053234796e-05, + "loss": 0.013, + "step": 6040 + }, + { + "epoch": 10.134003350083752, + "grad_norm": 0.07623566687107086, + "learning_rate": 1.6036725145226774e-05, + "loss": 0.0138, + "step": 6050 + }, + { + "epoch": 10.150753768844222, + "grad_norm": 0.0815720483660698, + "learning_rate": 1.5995483363980394e-05, + "loss": 0.0138, + "step": 6060 + }, + { + "epoch": 10.16750418760469, + "grad_norm": 0.061299752444028854, + "learning_rate": 1.595423402269956e-05, + "loss": 0.0137, + "step": 6070 + }, + { + "epoch": 10.18425460636516, + "grad_norm": 0.05670687556266785, + "learning_rate": 1.5912977434645608e-05, + "loss": 0.0165, + "step": 6080 + }, + { + "epoch": 10.201005025125628, + "grad_norm": 0.07310479134321213, + "learning_rate": 1.587171391313489e-05, + "loss": 0.0135, + "step": 6090 + }, + { + "epoch": 10.217755443886098, + "grad_norm": 0.06537134945392609, + "learning_rate": 1.5830443771536434e-05, + "loss": 0.0144, + "step": 6100 + }, + { + "epoch": 10.234505862646566, + "grad_norm": 0.07251140475273132, + "learning_rate": 1.578916732326953e-05, + "loss": 0.0157, + "step": 6110 + }, + { + "epoch": 10.251256281407034, + "grad_norm": 0.09199927002191544, + "learning_rate": 1.5747884881801364e-05, + "loss": 0.0144, + "step": 6120 + }, + { + "epoch": 10.268006700167504, + "grad_norm": 0.06975611299276352, + "learning_rate": 1.570659676064464e-05, + "loss": 0.0141, + "step": 6130 + }, + { + "epoch": 10.284757118927974, + "grad_norm": 0.0650688111782074, + "learning_rate": 1.5665303273355197e-05, + "loss": 0.0139, + "step": 6140 + }, + { + "epoch": 10.301507537688442, + "grad_norm": 0.11455053091049194, + "learning_rate": 1.562400473352962e-05, + "loss": 0.0134, + "step": 6150 + }, + { + "epoch": 10.318257956448912, + "grad_norm": 0.10255176573991776, + "learning_rate": 1.5582701454802867e-05, + "loss": 0.0126, + "step": 6160 + }, + { + "epoch": 10.33500837520938, + "grad_norm": 0.09504007548093796, + "learning_rate": 1.5541393750845886e-05, + "loss": 0.013, + "step": 6170 + }, + { + "epoch": 10.35175879396985, + "grad_norm": 0.053999241441488266, + "learning_rate": 1.5500081935363233e-05, + "loss": 0.0128, + "step": 6180 + }, + { + "epoch": 10.368509212730318, + "grad_norm": 0.07180154323577881, + "learning_rate": 1.5458766322090684e-05, + "loss": 0.0131, + "step": 6190 + }, + { + "epoch": 10.385259631490788, + "grad_norm": 0.07966731488704681, + "learning_rate": 1.5417447224792862e-05, + "loss": 0.0144, + "step": 6200 + }, + { + "epoch": 10.402010050251256, + "grad_norm": 0.08139584958553314, + "learning_rate": 1.5376124957260847e-05, + "loss": 0.0151, + "step": 6210 + }, + { + "epoch": 10.418760469011726, + "grad_norm": 0.06023263931274414, + "learning_rate": 1.5334799833309792e-05, + "loss": 0.0131, + "step": 6220 + }, + { + "epoch": 10.435510887772194, + "grad_norm": 0.09527294337749481, + "learning_rate": 1.529347216677654e-05, + "loss": 0.0134, + "step": 6230 + }, + { + "epoch": 10.452261306532664, + "grad_norm": 0.07697620987892151, + "learning_rate": 1.5252142271517259e-05, + "loss": 0.0139, + "step": 6240 + }, + { + "epoch": 10.469011725293132, + "grad_norm": 0.08630616217851639, + "learning_rate": 1.5210810461405019e-05, + "loss": 0.0127, + "step": 6250 + }, + { + "epoch": 10.485762144053602, + "grad_norm": 0.08324746787548065, + "learning_rate": 1.5169477050327457e-05, + "loss": 0.0137, + "step": 6260 + }, + { + "epoch": 10.50251256281407, + "grad_norm": 0.10764531046152115, + "learning_rate": 1.5128142352184346e-05, + "loss": 0.0116, + "step": 6270 + }, + { + "epoch": 10.519262981574538, + "grad_norm": 0.09146895259618759, + "learning_rate": 1.5086806680885245e-05, + "loss": 0.0125, + "step": 6280 + }, + { + "epoch": 10.536013400335008, + "grad_norm": 0.06823863834142685, + "learning_rate": 1.5045470350347106e-05, + "loss": 0.0116, + "step": 6290 + }, + { + "epoch": 10.552763819095478, + "grad_norm": 0.08591805398464203, + "learning_rate": 1.5004133674491876e-05, + "loss": 0.0125, + "step": 6300 + }, + { + "epoch": 10.569514237855946, + "grad_norm": 0.09628190100193024, + "learning_rate": 1.4962796967244137e-05, + "loss": 0.014, + "step": 6310 + }, + { + "epoch": 10.586264656616414, + "grad_norm": 0.09630268812179565, + "learning_rate": 1.4921460542528706e-05, + "loss": 0.0141, + "step": 6320 + }, + { + "epoch": 10.603015075376884, + "grad_norm": 0.06456310302019119, + "learning_rate": 1.4880124714268248e-05, + "loss": 0.0129, + "step": 6330 + }, + { + "epoch": 10.619765494137354, + "grad_norm": 0.06395621597766876, + "learning_rate": 1.4838789796380901e-05, + "loss": 0.0137, + "step": 6340 + }, + { + "epoch": 10.636515912897822, + "grad_norm": 0.06975158303976059, + "learning_rate": 1.4797456102777899e-05, + "loss": 0.0159, + "step": 6350 + }, + { + "epoch": 10.653266331658292, + "grad_norm": 0.07676661759614944, + "learning_rate": 1.4756123947361164e-05, + "loss": 0.0129, + "step": 6360 + }, + { + "epoch": 10.67001675041876, + "grad_norm": 0.07465475797653198, + "learning_rate": 1.4714793644020942e-05, + "loss": 0.013, + "step": 6370 + }, + { + "epoch": 10.68676716917923, + "grad_norm": 0.08438120037317276, + "learning_rate": 1.4673465506633417e-05, + "loss": 0.0148, + "step": 6380 + }, + { + "epoch": 10.703517587939698, + "grad_norm": 0.0682619959115982, + "learning_rate": 1.4632139849058328e-05, + "loss": 0.0132, + "step": 6390 + }, + { + "epoch": 10.720268006700168, + "grad_norm": 0.05983544513583183, + "learning_rate": 1.4590816985136563e-05, + "loss": 0.0153, + "step": 6400 + }, + { + "epoch": 10.737018425460636, + "grad_norm": 0.06861865520477295, + "learning_rate": 1.4549497228687812e-05, + "loss": 0.0141, + "step": 6410 + }, + { + "epoch": 10.753768844221106, + "grad_norm": 0.07951993495225906, + "learning_rate": 1.4508180893508157e-05, + "loss": 0.0145, + "step": 6420 + }, + { + "epoch": 10.770519262981574, + "grad_norm": 0.09159310907125473, + "learning_rate": 1.4466868293367708e-05, + "loss": 0.0128, + "step": 6430 + }, + { + "epoch": 10.787269681742044, + "grad_norm": 0.06889096647500992, + "learning_rate": 1.4425559742008196e-05, + "loss": 0.0144, + "step": 6440 + }, + { + "epoch": 10.804020100502512, + "grad_norm": 0.07343833148479462, + "learning_rate": 1.438425555314061e-05, + "loss": 0.0133, + "step": 6450 + }, + { + "epoch": 10.820770519262982, + "grad_norm": 0.08436884731054306, + "learning_rate": 1.4342956040442816e-05, + "loss": 0.0128, + "step": 6460 + }, + { + "epoch": 10.83752093802345, + "grad_norm": 0.0685647502541542, + "learning_rate": 1.4301661517557155e-05, + "loss": 0.0134, + "step": 6470 + }, + { + "epoch": 10.85427135678392, + "grad_norm": 0.07545416057109833, + "learning_rate": 1.426037229808808e-05, + "loss": 0.0125, + "step": 6480 + }, + { + "epoch": 10.871021775544389, + "grad_norm": 0.06606476753950119, + "learning_rate": 1.4219088695599773e-05, + "loss": 0.0131, + "step": 6490 + }, + { + "epoch": 10.887772194304858, + "grad_norm": 0.06588952988386154, + "learning_rate": 1.4177811023613749e-05, + "loss": 0.012, + "step": 6500 + }, + { + "epoch": 10.904522613065327, + "grad_norm": 0.0672588050365448, + "learning_rate": 1.4136539595606493e-05, + "loss": 0.0141, + "step": 6510 + }, + { + "epoch": 10.921273031825796, + "grad_norm": 0.06932391226291656, + "learning_rate": 1.4095274725007064e-05, + "loss": 0.0135, + "step": 6520 + }, + { + "epoch": 10.938023450586265, + "grad_norm": 0.055177588015794754, + "learning_rate": 1.4054016725194727e-05, + "loss": 0.0118, + "step": 6530 + }, + { + "epoch": 10.954773869346734, + "grad_norm": 0.06075577810406685, + "learning_rate": 1.4012765909496567e-05, + "loss": 0.0118, + "step": 6540 + }, + { + "epoch": 10.971524288107203, + "grad_norm": 0.08867711573839188, + "learning_rate": 1.3971522591185111e-05, + "loss": 0.0141, + "step": 6550 + }, + { + "epoch": 10.988274706867672, + "grad_norm": 0.06817186623811722, + "learning_rate": 1.3930287083475943e-05, + "loss": 0.0121, + "step": 6560 + }, + { + "epoch": 11.00502512562814, + "grad_norm": 0.06277189403772354, + "learning_rate": 1.3889059699525338e-05, + "loss": 0.0142, + "step": 6570 + }, + { + "epoch": 11.02177554438861, + "grad_norm": 0.05601593852043152, + "learning_rate": 1.3847840752427874e-05, + "loss": 0.0125, + "step": 6580 + }, + { + "epoch": 11.038525963149079, + "grad_norm": 0.14753024280071259, + "learning_rate": 1.380663055521405e-05, + "loss": 0.0133, + "step": 6590 + }, + { + "epoch": 11.055276381909549, + "grad_norm": 0.07654228061437607, + "learning_rate": 1.3765429420847928e-05, + "loss": 0.0126, + "step": 6600 + }, + { + "epoch": 11.072026800670017, + "grad_norm": 0.08025944232940674, + "learning_rate": 1.3724237662224738e-05, + "loss": 0.0146, + "step": 6610 + }, + { + "epoch": 11.088777219430487, + "grad_norm": 0.06721494346857071, + "learning_rate": 1.3683055592168497e-05, + "loss": 0.0159, + "step": 6620 + }, + { + "epoch": 11.105527638190955, + "grad_norm": 0.058076489716768265, + "learning_rate": 1.3641883523429665e-05, + "loss": 0.0145, + "step": 6630 + }, + { + "epoch": 11.122278056951425, + "grad_norm": 0.10511145740747452, + "learning_rate": 1.3600721768682729e-05, + "loss": 0.016, + "step": 6640 + }, + { + "epoch": 11.139028475711893, + "grad_norm": 0.07932860404253006, + "learning_rate": 1.3559570640523858e-05, + "loss": 0.0148, + "step": 6650 + }, + { + "epoch": 11.155778894472363, + "grad_norm": 0.1065361350774765, + "learning_rate": 1.351843045146852e-05, + "loss": 0.013, + "step": 6660 + }, + { + "epoch": 11.17252931323283, + "grad_norm": 0.09081228077411652, + "learning_rate": 1.3477301513949103e-05, + "loss": 0.0141, + "step": 6670 + }, + { + "epoch": 11.1892797319933, + "grad_norm": 0.06857980042695999, + "learning_rate": 1.3436184140312546e-05, + "loss": 0.0114, + "step": 6680 + }, + { + "epoch": 11.206030150753769, + "grad_norm": 0.06644202023744583, + "learning_rate": 1.3395078642817976e-05, + "loss": 0.0115, + "step": 6690 + }, + { + "epoch": 11.222780569514239, + "grad_norm": 0.07019989937543869, + "learning_rate": 1.3353985333634321e-05, + "loss": 0.0134, + "step": 6700 + }, + { + "epoch": 11.239530988274707, + "grad_norm": 0.06523793935775757, + "learning_rate": 1.3312904524837948e-05, + "loss": 0.0135, + "step": 6710 + }, + { + "epoch": 11.256281407035177, + "grad_norm": 0.08410218358039856, + "learning_rate": 1.3271836528410299e-05, + "loss": 0.0147, + "step": 6720 + }, + { + "epoch": 11.273031825795645, + "grad_norm": 0.10058104246854782, + "learning_rate": 1.3230781656235505e-05, + "loss": 0.015, + "step": 6730 + }, + { + "epoch": 11.289782244556115, + "grad_norm": 0.053105153143405914, + "learning_rate": 1.3189740220098026e-05, + "loss": 0.0118, + "step": 6740 + }, + { + "epoch": 11.306532663316583, + "grad_norm": 0.06775388866662979, + "learning_rate": 1.31487125316803e-05, + "loss": 0.0161, + "step": 6750 + }, + { + "epoch": 11.323283082077053, + "grad_norm": 0.06501705944538116, + "learning_rate": 1.3107698902560344e-05, + "loss": 0.0153, + "step": 6760 + }, + { + "epoch": 11.34003350083752, + "grad_norm": 0.08275182545185089, + "learning_rate": 1.3066699644209404e-05, + "loss": 0.0116, + "step": 6770 + }, + { + "epoch": 11.35678391959799, + "grad_norm": 0.06643733382225037, + "learning_rate": 1.3025715067989606e-05, + "loss": 0.0155, + "step": 6780 + }, + { + "epoch": 11.373534338358459, + "grad_norm": 0.08800774067640305, + "learning_rate": 1.2984745485151559e-05, + "loss": 0.0141, + "step": 6790 + }, + { + "epoch": 11.390284757118929, + "grad_norm": 0.10282844305038452, + "learning_rate": 1.2943791206832008e-05, + "loss": 0.0144, + "step": 6800 + }, + { + "epoch": 11.407035175879397, + "grad_norm": 0.08494739234447479, + "learning_rate": 1.290285254405148e-05, + "loss": 0.0138, + "step": 6810 + }, + { + "epoch": 11.423785594639867, + "grad_norm": 0.07839672267436981, + "learning_rate": 1.2861929807711901e-05, + "loss": 0.0148, + "step": 6820 + }, + { + "epoch": 11.440536013400335, + "grad_norm": 0.06659016013145447, + "learning_rate": 1.2821023308594256e-05, + "loss": 0.0144, + "step": 6830 + }, + { + "epoch": 11.457286432160805, + "grad_norm": 0.10885549336671829, + "learning_rate": 1.2780133357356218e-05, + "loss": 0.013, + "step": 6840 + }, + { + "epoch": 11.474036850921273, + "grad_norm": 0.07157129049301147, + "learning_rate": 1.273926026452978e-05, + "loss": 0.0138, + "step": 6850 + }, + { + "epoch": 11.490787269681743, + "grad_norm": 0.05953820422291756, + "learning_rate": 1.2698404340518922e-05, + "loss": 0.0129, + "step": 6860 + }, + { + "epoch": 11.507537688442211, + "grad_norm": 0.03799299895763397, + "learning_rate": 1.2657565895597222e-05, + "loss": 0.0106, + "step": 6870 + }, + { + "epoch": 11.52428810720268, + "grad_norm": 0.0774364024400711, + "learning_rate": 1.2616745239905527e-05, + "loss": 0.0138, + "step": 6880 + }, + { + "epoch": 11.541038525963149, + "grad_norm": 0.07412974536418915, + "learning_rate": 1.2575942683449585e-05, + "loss": 0.0134, + "step": 6890 + }, + { + "epoch": 11.557788944723619, + "grad_norm": 0.07800350338220596, + "learning_rate": 1.2535158536097691e-05, + "loss": 0.014, + "step": 6900 + }, + { + "epoch": 11.574539363484087, + "grad_norm": 0.08868993818759918, + "learning_rate": 1.249439310757833e-05, + "loss": 0.0139, + "step": 6910 + }, + { + "epoch": 11.591289782244557, + "grad_norm": 0.07635314017534256, + "learning_rate": 1.2453646707477843e-05, + "loss": 0.0134, + "step": 6920 + }, + { + "epoch": 11.608040201005025, + "grad_norm": 0.08172111213207245, + "learning_rate": 1.2412919645238049e-05, + "loss": 0.0124, + "step": 6930 + }, + { + "epoch": 11.624790619765495, + "grad_norm": 0.06785856932401657, + "learning_rate": 1.2372212230153911e-05, + "loss": 0.0118, + "step": 6940 + }, + { + "epoch": 11.641541038525963, + "grad_norm": 0.07923078536987305, + "learning_rate": 1.2331524771371197e-05, + "loss": 0.0131, + "step": 6950 + }, + { + "epoch": 11.658291457286433, + "grad_norm": 0.05836864933371544, + "learning_rate": 1.2290857577884109e-05, + "loss": 0.0126, + "step": 6960 + }, + { + "epoch": 11.675041876046901, + "grad_norm": 0.07780711352825165, + "learning_rate": 1.2250210958532943e-05, + "loss": 0.0141, + "step": 6970 + }, + { + "epoch": 11.691792294807371, + "grad_norm": 0.08186667412519455, + "learning_rate": 1.2209585222001762e-05, + "loss": 0.0129, + "step": 6980 + }, + { + "epoch": 11.708542713567839, + "grad_norm": 0.08738984167575836, + "learning_rate": 1.216898067681603e-05, + "loss": 0.0129, + "step": 6990 + }, + { + "epoch": 11.725293132328309, + "grad_norm": 0.07625941932201385, + "learning_rate": 1.2128397631340274e-05, + "loss": 0.0127, + "step": 7000 + }, + { + "epoch": 11.742043551088777, + "grad_norm": 0.06589437276124954, + "learning_rate": 1.2087836393775756e-05, + "loss": 0.0126, + "step": 7010 + }, + { + "epoch": 11.758793969849247, + "grad_norm": 0.06688161194324493, + "learning_rate": 1.2047297272158112e-05, + "loss": 0.0129, + "step": 7020 + }, + { + "epoch": 11.775544388609715, + "grad_norm": 0.08182154595851898, + "learning_rate": 1.2006780574355017e-05, + "loss": 0.0127, + "step": 7030 + }, + { + "epoch": 11.792294807370185, + "grad_norm": 0.06617710739374161, + "learning_rate": 1.1966286608063866e-05, + "loss": 0.0145, + "step": 7040 + }, + { + "epoch": 11.809045226130653, + "grad_norm": 0.060059934854507446, + "learning_rate": 1.1925815680809419e-05, + "loss": 0.0114, + "step": 7050 + }, + { + "epoch": 11.825795644891123, + "grad_norm": 0.08330146223306656, + "learning_rate": 1.1885368099941459e-05, + "loss": 0.0127, + "step": 7060 + }, + { + "epoch": 11.842546063651591, + "grad_norm": 0.08132729679346085, + "learning_rate": 1.1844944172632485e-05, + "loss": 0.0133, + "step": 7070 + }, + { + "epoch": 11.859296482412061, + "grad_norm": 0.08308562636375427, + "learning_rate": 1.1804544205875353e-05, + "loss": 0.0136, + "step": 7080 + }, + { + "epoch": 11.87604690117253, + "grad_norm": 0.07975098490715027, + "learning_rate": 1.1764168506480953e-05, + "loss": 0.0132, + "step": 7090 + }, + { + "epoch": 11.892797319932999, + "grad_norm": 0.07059597223997116, + "learning_rate": 1.1723817381075889e-05, + "loss": 0.0126, + "step": 7100 + }, + { + "epoch": 11.909547738693467, + "grad_norm": 0.07069030404090881, + "learning_rate": 1.1683491136100135e-05, + "loss": 0.0127, + "step": 7110 + }, + { + "epoch": 11.926298157453937, + "grad_norm": 0.06913162767887115, + "learning_rate": 1.1643190077804713e-05, + "loss": 0.0124, + "step": 7120 + }, + { + "epoch": 11.943048576214405, + "grad_norm": 0.07565716654062271, + "learning_rate": 1.1602914512249379e-05, + "loss": 0.0139, + "step": 7130 + }, + { + "epoch": 11.959798994974875, + "grad_norm": 0.05990506708621979, + "learning_rate": 1.156266474530028e-05, + "loss": 0.0111, + "step": 7140 + }, + { + "epoch": 11.976549413735343, + "grad_norm": 0.0784129723906517, + "learning_rate": 1.152244108262764e-05, + "loss": 0.0099, + "step": 7150 + }, + { + "epoch": 11.993299832495813, + "grad_norm": 0.0859949141740799, + "learning_rate": 1.1482243829703444e-05, + "loss": 0.0129, + "step": 7160 + }, + { + "epoch": 12.010050251256281, + "grad_norm": 0.0738619789481163, + "learning_rate": 1.1442073291799106e-05, + "loss": 0.0122, + "step": 7170 + }, + { + "epoch": 12.026800670016751, + "grad_norm": 0.09288010001182556, + "learning_rate": 1.1401929773983155e-05, + "loss": 0.0134, + "step": 7180 + }, + { + "epoch": 12.04355108877722, + "grad_norm": 0.07321544736623764, + "learning_rate": 1.1361813581118936e-05, + "loss": 0.0145, + "step": 7190 + }, + { + "epoch": 12.06030150753769, + "grad_norm": 0.07173691689968109, + "learning_rate": 1.1321725017862268e-05, + "loss": 0.0115, + "step": 7200 + }, + { + "epoch": 12.077051926298157, + "grad_norm": 0.058314502239227295, + "learning_rate": 1.1281664388659134e-05, + "loss": 0.0128, + "step": 7210 + }, + { + "epoch": 12.093802345058627, + "grad_norm": 0.07083640247583389, + "learning_rate": 1.1241631997743384e-05, + "loss": 0.0101, + "step": 7220 + }, + { + "epoch": 12.110552763819095, + "grad_norm": 0.065085269510746, + "learning_rate": 1.1201628149134422e-05, + "loss": 0.0137, + "step": 7230 + }, + { + "epoch": 12.127303182579565, + "grad_norm": 0.0781814306974411, + "learning_rate": 1.116165314663489e-05, + "loss": 0.0146, + "step": 7240 + }, + { + "epoch": 12.144053601340033, + "grad_norm": 0.09068423509597778, + "learning_rate": 1.1121707293828353e-05, + "loss": 0.014, + "step": 7250 + }, + { + "epoch": 12.160804020100503, + "grad_norm": 0.08548370748758316, + "learning_rate": 1.1081790894077022e-05, + "loss": 0.0121, + "step": 7260 + }, + { + "epoch": 12.177554438860971, + "grad_norm": 0.07648133486509323, + "learning_rate": 1.104190425051941e-05, + "loss": 0.0131, + "step": 7270 + }, + { + "epoch": 12.194304857621441, + "grad_norm": 0.09778308868408203, + "learning_rate": 1.1002047666068066e-05, + "loss": 0.0139, + "step": 7280 + }, + { + "epoch": 12.21105527638191, + "grad_norm": 0.07223774492740631, + "learning_rate": 1.0962221443407257e-05, + "loss": 0.0137, + "step": 7290 + }, + { + "epoch": 12.22780569514238, + "grad_norm": 0.0755016952753067, + "learning_rate": 1.0922425884990673e-05, + "loss": 0.0119, + "step": 7300 + }, + { + "epoch": 12.244556113902847, + "grad_norm": 0.0679793432354927, + "learning_rate": 1.0882661293039124e-05, + "loss": 0.0127, + "step": 7310 + }, + { + "epoch": 12.261306532663317, + "grad_norm": 0.08829930424690247, + "learning_rate": 1.0842927969538258e-05, + "loss": 0.0141, + "step": 7320 + }, + { + "epoch": 12.278056951423785, + "grad_norm": 0.06518194079399109, + "learning_rate": 1.0803226216236256e-05, + "loss": 0.0125, + "step": 7330 + }, + { + "epoch": 12.294807370184255, + "grad_norm": 0.08196015655994415, + "learning_rate": 1.076355633464154e-05, + "loss": 0.0114, + "step": 7340 + }, + { + "epoch": 12.311557788944723, + "grad_norm": 0.06805971264839172, + "learning_rate": 1.0723918626020498e-05, + "loss": 0.0118, + "step": 7350 + }, + { + "epoch": 12.328308207705193, + "grad_norm": 0.0839708149433136, + "learning_rate": 1.0684313391395176e-05, + "loss": 0.0148, + "step": 7360 + }, + { + "epoch": 12.345058626465661, + "grad_norm": 0.07113301753997803, + "learning_rate": 1.0644740931541005e-05, + "loss": 0.015, + "step": 7370 + }, + { + "epoch": 12.361809045226131, + "grad_norm": 0.08832506835460663, + "learning_rate": 1.0605201546984517e-05, + "loss": 0.0148, + "step": 7380 + }, + { + "epoch": 12.3785594639866, + "grad_norm": 0.09089552611112595, + "learning_rate": 1.0565695538001056e-05, + "loss": 0.016, + "step": 7390 + }, + { + "epoch": 12.39530988274707, + "grad_norm": 0.06780517846345901, + "learning_rate": 1.0526223204612495e-05, + "loss": 0.0119, + "step": 7400 + }, + { + "epoch": 12.412060301507537, + "grad_norm": 0.07529979944229126, + "learning_rate": 1.0486784846584974e-05, + "loss": 0.0111, + "step": 7410 + }, + { + "epoch": 12.428810720268007, + "grad_norm": 0.08596140146255493, + "learning_rate": 1.0447380763426607e-05, + "loss": 0.0122, + "step": 7420 + }, + { + "epoch": 12.445561139028475, + "grad_norm": 0.06264451146125793, + "learning_rate": 1.040801125438521e-05, + "loss": 0.013, + "step": 7430 + }, + { + "epoch": 12.462311557788945, + "grad_norm": 0.06601212918758392, + "learning_rate": 1.0368676618446042e-05, + "loss": 0.0132, + "step": 7440 + }, + { + "epoch": 12.479061976549414, + "grad_norm": 0.08287645131349564, + "learning_rate": 1.0329377154329512e-05, + "loss": 0.0127, + "step": 7450 + }, + { + "epoch": 12.495812395309883, + "grad_norm": 0.10921191424131393, + "learning_rate": 1.0290113160488925e-05, + "loss": 0.0125, + "step": 7460 + }, + { + "epoch": 12.512562814070352, + "grad_norm": 0.07738841325044632, + "learning_rate": 1.0250884935108223e-05, + "loss": 0.0142, + "step": 7470 + }, + { + "epoch": 12.529313232830821, + "grad_norm": 0.07163051515817642, + "learning_rate": 1.0211692776099698e-05, + "loss": 0.0118, + "step": 7480 + }, + { + "epoch": 12.54606365159129, + "grad_norm": 0.07590950280427933, + "learning_rate": 1.0172536981101747e-05, + "loss": 0.0093, + "step": 7490 + }, + { + "epoch": 12.56281407035176, + "grad_norm": 0.08660171180963516, + "learning_rate": 1.0133417847476606e-05, + "loss": 0.012, + "step": 7500 + }, + { + "epoch": 12.579564489112228, + "grad_norm": 0.07874178141355515, + "learning_rate": 1.0094335672308095e-05, + "loss": 0.0117, + "step": 7510 + }, + { + "epoch": 12.596314907872697, + "grad_norm": 0.07014073431491852, + "learning_rate": 1.0055290752399353e-05, + "loss": 0.0138, + "step": 7520 + }, + { + "epoch": 12.613065326633166, + "grad_norm": 0.07678911834955215, + "learning_rate": 1.0016283384270594e-05, + "loss": 0.0117, + "step": 7530 + }, + { + "epoch": 12.629815745393635, + "grad_norm": 0.06955492496490479, + "learning_rate": 9.977313864156854e-06, + "loss": 0.0108, + "step": 7540 + }, + { + "epoch": 12.646566164154104, + "grad_norm": 0.08744408190250397, + "learning_rate": 9.938382488005729e-06, + "loss": 0.0116, + "step": 7550 + }, + { + "epoch": 12.663316582914574, + "grad_norm": 0.08456790447235107, + "learning_rate": 9.89948955147515e-06, + "loss": 0.0142, + "step": 7560 + }, + { + "epoch": 12.680067001675042, + "grad_norm": 0.08989357948303223, + "learning_rate": 9.860635349931116e-06, + "loss": 0.011, + "step": 7570 + }, + { + "epoch": 12.696817420435512, + "grad_norm": 0.05640851706266403, + "learning_rate": 9.821820178445455e-06, + "loss": 0.0116, + "step": 7580 + }, + { + "epoch": 12.71356783919598, + "grad_norm": 0.07872806489467621, + "learning_rate": 9.783044331793605e-06, + "loss": 0.0106, + "step": 7590 + }, + { + "epoch": 12.73031825795645, + "grad_norm": 0.08626449853181839, + "learning_rate": 9.744308104452338e-06, + "loss": 0.0118, + "step": 7600 + }, + { + "epoch": 12.747068676716918, + "grad_norm": 0.05731945484876633, + "learning_rate": 9.705611790597562e-06, + "loss": 0.0114, + "step": 7610 + }, + { + "epoch": 12.763819095477388, + "grad_norm": 0.06404105573892593, + "learning_rate": 9.666955684102043e-06, + "loss": 0.0123, + "step": 7620 + }, + { + "epoch": 12.780569514237856, + "grad_norm": 0.0529644712805748, + "learning_rate": 9.62834007853323e-06, + "loss": 0.0127, + "step": 7630 + }, + { + "epoch": 12.797319932998326, + "grad_norm": 0.10183175653219223, + "learning_rate": 9.589765267150975e-06, + "loss": 0.0123, + "step": 7640 + }, + { + "epoch": 12.814070351758794, + "grad_norm": 0.061799515038728714, + "learning_rate": 9.551231542905326e-06, + "loss": 0.0134, + "step": 7650 + }, + { + "epoch": 12.830820770519264, + "grad_norm": 0.04345851019024849, + "learning_rate": 9.512739198434312e-06, + "loss": 0.0137, + "step": 7660 + }, + { + "epoch": 12.847571189279732, + "grad_norm": 0.057996947318315506, + "learning_rate": 9.474288526061698e-06, + "loss": 0.0144, + "step": 7670 + }, + { + "epoch": 12.864321608040202, + "grad_norm": 0.07847338914871216, + "learning_rate": 9.435879817794784e-06, + "loss": 0.0123, + "step": 7680 + }, + { + "epoch": 12.88107202680067, + "grad_norm": 0.0713241770863533, + "learning_rate": 9.39751336532218e-06, + "loss": 0.0101, + "step": 7690 + }, + { + "epoch": 12.89782244556114, + "grad_norm": 0.07546516507863998, + "learning_rate": 9.359189460011591e-06, + "loss": 0.0139, + "step": 7700 + }, + { + "epoch": 12.914572864321608, + "grad_norm": 0.04970499873161316, + "learning_rate": 9.3209083929076e-06, + "loss": 0.0142, + "step": 7710 + }, + { + "epoch": 12.931323283082078, + "grad_norm": 0.08262616395950317, + "learning_rate": 9.282670454729471e-06, + "loss": 0.0115, + "step": 7720 + }, + { + "epoch": 12.948073701842546, + "grad_norm": 0.055856745690107346, + "learning_rate": 9.244475935868927e-06, + "loss": 0.0117, + "step": 7730 + }, + { + "epoch": 12.964824120603016, + "grad_norm": 0.04862665757536888, + "learning_rate": 9.206325126387945e-06, + "loss": 0.0104, + "step": 7740 + }, + { + "epoch": 12.981574539363484, + "grad_norm": 0.06921498477458954, + "learning_rate": 9.168218316016572e-06, + "loss": 0.0123, + "step": 7750 + }, + { + "epoch": 12.998324958123954, + "grad_norm": 0.09353175759315491, + "learning_rate": 9.130155794150698e-06, + "loss": 0.0121, + "step": 7760 + }, + { + "epoch": 13.015075376884422, + "grad_norm": 0.07180416584014893, + "learning_rate": 9.092137849849876e-06, + "loss": 0.0133, + "step": 7770 + }, + { + "epoch": 13.031825795644892, + "grad_norm": 0.07870851457118988, + "learning_rate": 9.054164771835128e-06, + "loss": 0.0123, + "step": 7780 + }, + { + "epoch": 13.04857621440536, + "grad_norm": 0.06702952831983566, + "learning_rate": 9.016236848486736e-06, + "loss": 0.0107, + "step": 7790 + }, + { + "epoch": 13.06532663316583, + "grad_norm": 0.07762094587087631, + "learning_rate": 8.978354367842068e-06, + "loss": 0.011, + "step": 7800 + }, + { + "epoch": 13.082077051926298, + "grad_norm": 0.07945635914802551, + "learning_rate": 8.940517617593384e-06, + "loss": 0.0119, + "step": 7810 + }, + { + "epoch": 13.098827470686768, + "grad_norm": 0.06439702212810516, + "learning_rate": 8.902726885085655e-06, + "loss": 0.0095, + "step": 7820 + }, + { + "epoch": 13.115577889447236, + "grad_norm": 0.0856538638472557, + "learning_rate": 8.864982457314365e-06, + "loss": 0.0142, + "step": 7830 + }, + { + "epoch": 13.132328308207706, + "grad_norm": 0.07062462717294693, + "learning_rate": 8.827284620923369e-06, + "loss": 0.0126, + "step": 7840 + }, + { + "epoch": 13.149078726968174, + "grad_norm": 0.08214748650789261, + "learning_rate": 8.789633662202667e-06, + "loss": 0.0122, + "step": 7850 + }, + { + "epoch": 13.165829145728644, + "grad_norm": 0.06878134608268738, + "learning_rate": 8.752029867086264e-06, + "loss": 0.0119, + "step": 7860 + }, + { + "epoch": 13.182579564489112, + "grad_norm": 0.10038811713457108, + "learning_rate": 8.714473521149994e-06, + "loss": 0.012, + "step": 7870 + }, + { + "epoch": 13.199329983249582, + "grad_norm": 0.05590192601084709, + "learning_rate": 8.676964909609334e-06, + "loss": 0.0121, + "step": 7880 + }, + { + "epoch": 13.21608040201005, + "grad_norm": 0.08674315363168716, + "learning_rate": 8.639504317317258e-06, + "loss": 0.013, + "step": 7890 + }, + { + "epoch": 13.23283082077052, + "grad_norm": 0.07082588225603104, + "learning_rate": 8.602092028762073e-06, + "loss": 0.0133, + "step": 7900 + }, + { + "epoch": 13.249581239530988, + "grad_norm": 0.09573926776647568, + "learning_rate": 8.56472832806523e-06, + "loss": 0.0137, + "step": 7910 + }, + { + "epoch": 13.266331658291458, + "grad_norm": 0.073761947453022, + "learning_rate": 8.527413498979204e-06, + "loss": 0.0112, + "step": 7920 + }, + { + "epoch": 13.283082077051926, + "grad_norm": 0.06695816665887833, + "learning_rate": 8.490147824885328e-06, + "loss": 0.0112, + "step": 7930 + }, + { + "epoch": 13.299832495812396, + "grad_norm": 0.09559398889541626, + "learning_rate": 8.452931588791614e-06, + "loss": 0.0126, + "step": 7940 + }, + { + "epoch": 13.316582914572864, + "grad_norm": 0.0765177384018898, + "learning_rate": 8.415765073330642e-06, + "loss": 0.0121, + "step": 7950 + }, + { + "epoch": 13.333333333333334, + "grad_norm": 0.06615452468395233, + "learning_rate": 8.378648560757396e-06, + "loss": 0.0131, + "step": 7960 + }, + { + "epoch": 13.350083752093802, + "grad_norm": 0.08720266819000244, + "learning_rate": 8.34158233294711e-06, + "loss": 0.0135, + "step": 7970 + }, + { + "epoch": 13.366834170854272, + "grad_norm": 0.07875142246484756, + "learning_rate": 8.304566671393151e-06, + "loss": 0.0126, + "step": 7980 + }, + { + "epoch": 13.38358458961474, + "grad_norm": 0.04822768643498421, + "learning_rate": 8.267601857204869e-06, + "loss": 0.0119, + "step": 7990 + }, + { + "epoch": 13.40033500837521, + "grad_norm": 0.062463391572237015, + "learning_rate": 8.230688171105446e-06, + "loss": 0.011, + "step": 8000 + } + ], + "logging_steps": 10, + "max_steps": 12000, + "num_input_tokens_seen": 0, + "num_train_epochs": 21, + "save_steps": 12001, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/config.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/config.json new file mode 100644 index 0000000000000000000000000000000000000000..47d7b77e4255263f3803a4b6f94ca6831e459b70 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/config.json @@ -0,0 +1,64 @@ +{ + "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" + ], + "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", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_no_video/checkpoint-5000/eval.log b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_no_video/checkpoint-5000/eval.log new file mode 100644 index 0000000000000000000000000000000000000000..da390f2a207d03332996129df7744ed795b82702 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_no_video/checkpoint-5000/eval.log @@ -0,0 +1,68 @@ +[Thu Jul 2 10:03:33 KST 2026] START gpu=2 port=18751 ckpt=5000 +model_path=/home/seonho/groot_robocasa/robocasa_v2/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-5000 +video_dir=/home/seonho/groot_robocasa/robocasa_v2/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_no_video/checkpoint-5000 +[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. +WARNING: mimicgen environments not imported since mimicgen is not installed! +Model not found or avail in the huggingface hub. Loading from local path: /home/seonho/groot_robocasa/robocasa_v2/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-5000 +Loading pretrained dual brain from /home/seonho/groot_robocasa/robocasa_v2/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-5000 +Tune backbone vision tower: True +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/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/checkpoint-5000 +Available modality configs: +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 +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/vector/async_vector_env.py:422: UserWarning: ERROR: Received the following error from Worker-7 - Shutting it down + self._raise_if_errors(successes) +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/vector/async_vector_env.py:422: UserWarning: ERROR: Traceback (most recent call last): + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/vector/async_vector_env.py", line 714, in _async_worker + observation, info = env.reset() + ^^^^^^^^^^^ + File "/home/seonho/clvla/benchmarks/robocasa_v2/Isaac-GR00T/gr00t/eval/wrappers/multistep_wrapper.py", line 189, in reset + obs, info = super().reset(seed=seed, options=options) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/core.py", line 328, in reset + return self.env.reset(seed=seed, options=options) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/seonho/clvla/benchmarks/robocasa_v2/Isaac-GR00T/gr00t/eval/wrappers/video_recording_wrapper.py", line 259, in reset + result = super().reset(**kwargs) + ^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/core.py", line 328, in reset + return self.env.reset(seed=seed, options=options) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/seonho/clvla/benchmarks/robocasa_v2/my_scripts/Fewshot/fewshot_eval.py", line 133, in reset + obs, info = super().reset(seed=reset_seed, options=options) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/core.py", line 328, in reset + return self.env.reset(seed=seed, options=options) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/wrappers/common.py", line 400, in reset + return super().reset(seed=seed, options=options) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/core.py", line 328, in reset + return self.env.reset(seed=seed, options=options) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/wrappers/common.py", line 295, in reset + return self.env.reset(seed=seed, options=options) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/robocasa/wrappers/gym_wrapper.py", line 361, in reset + raw_obs = self.env.reset() + ^^^^^^^^^^^^^^^^ + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/environments/base.py", line 302, in reset + self._reset_internal() + File "/home/seonho/clvla/benchmarks/robocasa_v2/robocasa/robocasa/environments/kitchen/kitchen.py", line 1099, in _reset_internal + super()._reset_internal() + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/environments/robot_env.py", line 524, in _reset_internal + super()._reset_internal() + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/environments/base.py", line 380, in _reset_internal + render_context = MjRenderContextOffscreen(self.sim, device_id=self.render_gpu_device_id) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 218, in __init__ + super().__init__(sim, offscreen=True, device_id=device_id, max_width=max_width, max_height=max_height) + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 115, in __init__ + self._set_mujoco_context_and_buffers() + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 118, in _set_mujoco_context_and_buffers + self.con = mujoco.MjrContext(self.model._model, mujoco.mjtFontScale.mjFONTSCALE_150) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ +mujoco.FatalError: Offscreen framebuffer is not complete, error 0x8cdd + + self._raise_if_errors(successes) +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/vector/async_vector_env.py:422: UserWarning: ERROR: Raising the last exception back to the main process. + self._raise_if_errors(successes) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +Exception ignored in: +Traceback (most recent call last): + File "/home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/utils/binding_utils.py", line 202, in __del__ + self.con.free() + ^^^^^^^^ +AttributeError: 'MjRenderContextOffscreen' object has no attribute 'con' +EP 1 success: False; Cumulative success rate: 0.0 +EP 2 success: False; Cumulative success rate: 0.0 +EP 3 success: False; Cumulative success rate: 0.0 +EP 4 success: False; Cumulative success rate: 0.0 +EP 5 success: False; Cumulative success rate: 0.0 +EP 6 success: False; Cumulative success rate: 0.0 +EP 7 success: False; Cumulative success rate: 0.0 +EP 8 success: False; Cumulative success rate: 0.0 +EP 9 success: False; Cumulative success rate: 0.0 +EP 10 success: False; Cumulative success rate: 0.0 +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for CloseDishwasher... +Running 50 episodes for robocasa/CloseDishwasher with 10 environments +Creating CloseDishwasher with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for CloseDrawer... +Running 50 episodes for robocasa/CloseDrawer with 10 environments +Creating CloseDrawer with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for CloseElectricKettleLid... +Running 50 episodes for robocasa/CloseElectricKettleLid with 10 environments +Creating CloseElectricKettleLid with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for CloseFridgeDrawer... +Running 50 episodes for robocasa/CloseFridgeDrawer with 10 environments +Creating CloseFridgeDrawer with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for CloseMicrowave... +Running 50 episodes for robocasa/CloseMicrowave with 10 environments +Creating CloseMicrowave with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for CloseOven... +Running 50 episodes for robocasa/CloseOven with 10 environments +Creating CloseOven with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for CloseStandMixerHead... +Running 50 episodes for robocasa/CloseStandMixerHead with 10 environments +Creating CloseStandMixerHead with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for CoffeeServeMug... +Running 50 episodes for robocasa/CoffeeServeMug with 10 environments +Creating CoffeeServeMug with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for LowerHeat... +Running 50 episodes for robocasa/LowerHeat with 10 environments +Creating LowerHeat with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for MakeIcedCoffee... +Running 50 episodes for robocasa/MakeIcedCoffee with 10 environments +Creating MakeIcedCoffee with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for OpenBlenderLid... +Running 50 episodes for robocasa/OpenBlenderLid with 10 environments +Creating OpenBlenderLid with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for OpenDishwasher... +Running 50 episodes for robocasa/OpenDishwasher with 10 environments +Creating OpenDishwasher with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for OpenFridge... +Running 50 episodes for robocasa/OpenFridge with 10 environments +Creating OpenFridge with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for OpenFridgeDrawer... +Running 50 episodes for robocasa/OpenFridgeDrawer with 10 environments +Creating OpenFridgeDrawer with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for OpenMicrowave... +Running 50 episodes for robocasa/OpenMicrowave with 10 environments +Creating OpenMicrowave with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for OpenOven... +Running 50 episodes for robocasa/OpenOven with 10 environments +Creating OpenOven with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for OpenToasterOvenDoor... +Running 50 episodes for robocasa/OpenToasterOvenDoor with 10 environments +Creating OpenToasterOvenDoor with split=pretrain +Exception! Offscreen framebuffer is not complete, error 0x8cdd +Running simulation for PackDessert... +Running 50 episodes for robocasa/PackDessert with 10 environments +Creating PackDessert with split=pretrain +[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] 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] 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] 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] 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) +[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) +[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) +[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) +[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) +/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() +/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() +/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() +/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() +/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. +`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. +`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. +`use_fast` is set to `True` but the image processor class does not have a fast version. Falling back to the slow version. +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +`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. +`use_fast` is set to `True` but the image processor class does not have a fast version. Falling back to the slow version. +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +[robosuite INFO] Loading controller configuration from: /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/controllers/config/robots/default_pandaomron.json (composite_controller_factory.py:121) +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `reset()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `reset()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `reset()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `reset()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `reset()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `reset()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `reset()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `reset()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `reset()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `reset()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `step()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `step()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `step()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `step()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `step()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `step()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `step()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `step()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `step()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/gymnasium/utils/passive_env_checker.py:158: UserWarning: WARN: The obs returned by the `step()` method is not within the observation space. + logger.warn(f"{pre} is not within the observation space.") diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/eval_manifest.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/eval_manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..28b7f392b4a3705b1cf78a23971bb070af5672f8 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/eval_manifest.json @@ -0,0 +1,11939 @@ +{ + "version": 1, + "split": "pretrain", + "task_set": [ + "fewshot_heldout_atomic39" + ], + "n_episodes_per_task": 50, + "episode_seed_start": 42, + "tasks": { + "CheesyBread": [ + { + "task_name": "CheesyBread", + "episode_idx": 0, + "seed": 42, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 1, + "seed": 43, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 2, + "seed": 44, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 27.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 3, + "seed": 45, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 31.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 4, + "seed": 46, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 5, + "seed": 47, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 27.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 6, + "seed": 48, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 26.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 7, + "seed": 49, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 8, + "seed": 50, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 9, + "seed": 51, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 29.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 10, + "seed": 52, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 11, + "seed": 53, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 12, + "seed": 54, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 13, + "seed": 55, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 14, + "seed": 56, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 29.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 15, + "seed": 57, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 32.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 16, + "seed": 58, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 17, + "seed": 59, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 18, + "seed": 60, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 25.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 19, + "seed": 61, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 18.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 20, + "seed": 62, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 21, + "seed": 63, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 22, + "seed": 64, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 20.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 23, + "seed": 65, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 10.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 24, + "seed": 66, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 25, + "seed": 67, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 26, + "seed": 68, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 27, + "seed": 69, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 28, + "seed": 70, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 29, + "seed": 71, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 30, + "seed": 72, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 24.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 31, + "seed": 73, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 32, + "seed": 74, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 28.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 33, + "seed": 75, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 24.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 34, + "seed": 76, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 35, + "seed": 77, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 36, + "seed": 78, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 37, + "seed": 79, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 29.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 38, + "seed": 80, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 39, + "seed": 81, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 28.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 40, + "seed": 82, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 41, + "seed": 83, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 31.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 42, + "seed": 84, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 43, + "seed": 85, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 28.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 44, + "seed": 86, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 45, + "seed": 87, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 46, + "seed": 88, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 21.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 47, + "seed": 89, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 48, + "seed": 90, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 21.0 + }, + { + "task_name": "CheesyBread", + "episode_idx": 49, + "seed": 91, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0 + } + ], + "CloseCabinet": [ + { + "task_name": "CloseCabinet", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "CloseCabinet", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "CloseDishwasher": [ + { + "task_name": "CloseDishwasher", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "CloseDishwasher", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "CloseDrawer": [ + { + "task_name": "CloseDrawer", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "CloseDrawer", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "CloseElectricKettleLid": [ + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "CloseElectricKettleLid", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "CloseFridgeDrawer": [ + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "CloseFridgeDrawer", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "CloseMicrowave": [ + { + "task_name": "CloseMicrowave", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "CloseMicrowave", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "CloseOven": [ + { + "task_name": "CloseOven", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "CloseOven", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "CloseStandMixerHead": [ + { + "task_name": "CloseStandMixerHead", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "CloseStandMixerHead", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "CoffeeServeMug": [ + { + "task_name": "CoffeeServeMug", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "CoffeeServeMug", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "LowerHeat": [ + { + "task_name": "LowerHeat", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "LowerHeat", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "MakeIcedCoffee": [ + { + "task_name": "MakeIcedCoffee", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "MakeIcedCoffee", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "OpenBlenderLid": [ + { + "task_name": "OpenBlenderLid", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "OpenBlenderLid", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "OpenDishwasher": [ + { + "task_name": "OpenDishwasher", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "OpenDishwasher", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "OpenFridge": [ + { + "task_name": "OpenFridge", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "OpenFridge", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "OpenFridgeDrawer": [ + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "OpenFridgeDrawer", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "OpenMicrowave": [ + { + "task_name": "OpenMicrowave", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "OpenMicrowave", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "OpenOven": [ + { + "task_name": "OpenOven", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "OpenOven", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "OpenToasterOvenDoor": [ + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "OpenToasterOvenDoor", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "PackDessert": [ + { + "task_name": "PackDessert", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "PackDessert", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "PickPlaceCounterToBlender": [ + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "PickPlaceCounterToBlender", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "PickPlaceCounterToDrawer": [ + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "PickPlaceCounterToDrawer", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "PickPlaceCounterToMicrowave": [ + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "PickPlaceCounterToMicrowave", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "PickPlaceCounterToOven": [ + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "PickPlaceCounterToOven", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "PickPlaceCounterToSink": [ + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "PickPlaceCounterToSink", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "PickPlaceCounterToStandMixer": [ + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "PickPlaceCounterToStandMixer", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "PickPlaceCounterToToasterOven": [ + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "PickPlaceCounterToToasterOven", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "PickPlaceFridgeDrawerToShelf": [ + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "PickPlaceFridgeDrawerToShelf", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "PickPlaceFridgeShelfToDrawer": [ + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "PickPlaceFridgeShelfToDrawer", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "PickPlaceMicrowaveToCounter": [ + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "PickPlaceMicrowaveToCounter", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "PickPlaceStoveToCounter": [ + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "PickPlaceStoveToCounter", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "PickPlaceToasterOvenToCounter": [ + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "PickPlaceToasterOvenToCounter", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "PreheatOven": [ + { + "task_name": "PreheatOven", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "PreheatOven", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "SlideOvenRack": [ + { + "task_name": "SlideOvenRack", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "SlideOvenRack", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "SlideToasterOvenRack": [ + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "SlideToasterOvenRack", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "StartCoffeeMachine": [ + { + "task_name": "StartCoffeeMachine", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "StartCoffeeMachine", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "TurnOnBlender": [ + { + "task_name": "TurnOnBlender", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "TurnOnBlender", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "TurnOnToaster": [ + { + "task_name": "TurnOnToaster", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "TurnOnToaster", + "episode_idx": 49, + "seed": 91, + "description": null + } + ], + "TurnOnToasterOven": [ + { + "task_name": "TurnOnToasterOven", + "episode_idx": 0, + "seed": 42, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 1, + "seed": 43, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 2, + "seed": 44, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 3, + "seed": 45, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 4, + "seed": 46, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 5, + "seed": 47, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 6, + "seed": 48, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 7, + "seed": 49, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 8, + "seed": 50, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 9, + "seed": 51, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 10, + "seed": 52, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 11, + "seed": 53, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 12, + "seed": 54, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 13, + "seed": 55, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 14, + "seed": 56, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 15, + "seed": 57, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 16, + "seed": 58, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 17, + "seed": 59, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 18, + "seed": 60, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 19, + "seed": 61, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 20, + "seed": 62, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 21, + "seed": 63, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 22, + "seed": 64, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 23, + "seed": 65, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 24, + "seed": 66, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 25, + "seed": 67, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 26, + "seed": 68, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 27, + "seed": 69, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 28, + "seed": 70, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 29, + "seed": 71, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 30, + "seed": 72, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 31, + "seed": 73, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 32, + "seed": 74, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 33, + "seed": 75, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 34, + "seed": 76, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 35, + "seed": 77, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 36, + "seed": 78, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 37, + "seed": 79, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 38, + "seed": 80, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 39, + "seed": 81, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 40, + "seed": 82, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 41, + "seed": 83, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 42, + "seed": 84, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 43, + "seed": 85, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 44, + "seed": 86, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 45, + "seed": 87, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 46, + "seed": 88, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 47, + "seed": 89, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 48, + "seed": 90, + "description": null + }, + { + "task_name": "TurnOnToasterOven", + "episode_idx": 49, + "seed": 91, + "description": null + } + ] + } +} \ No newline at end of file diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/CheesyBread/episode_results.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/CheesyBread/episode_results.json new file mode 100644 index 0000000000000000000000000000000000000000..960696ad93a665c5e82a4bb4d28a3bd7d4fac627 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/CheesyBread/episode_results.json @@ -0,0 +1,602 @@ +[ + { + "completion_order": 0, + "task_name": "CheesyBread", + "episode_idx": 0, + "seed": 42, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 1, + "task_name": "CheesyBread", + "episode_idx": 1, + "seed": 43, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 2, + "task_name": "CheesyBread", + "episode_idx": 2, + "seed": 44, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 27.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 3, + "task_name": "CheesyBread", + "episode_idx": 3, + "seed": 45, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 31.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 4, + "task_name": "CheesyBread", + "episode_idx": 4, + "seed": 46, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 5, + "task_name": "CheesyBread", + "episode_idx": 5, + "seed": 47, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 27.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 6, + "task_name": "CheesyBread", + "episode_idx": 6, + "seed": 48, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 26.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 7, + "task_name": "CheesyBread", + "episode_idx": 7, + "seed": 49, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 8, + "task_name": "CheesyBread", + "episode_idx": 8, + "seed": 50, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 9, + "task_name": "CheesyBread", + "episode_idx": 9, + "seed": 51, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 29.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 10, + "task_name": "CheesyBread", + "episode_idx": 10, + "seed": 52, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 11, + "task_name": "CheesyBread", + "episode_idx": 11, + "seed": 53, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 12, + "task_name": "CheesyBread", + "episode_idx": 12, + "seed": 54, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 13, + "task_name": "CheesyBread", + "episode_idx": 13, + "seed": 55, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 14, + "task_name": "CheesyBread", + "episode_idx": 14, + "seed": 56, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 29.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 15, + "task_name": "CheesyBread", + "episode_idx": 15, + "seed": 57, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 32.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 16, + "task_name": "CheesyBread", + "episode_idx": 16, + "seed": 58, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 17, + "task_name": "CheesyBread", + "episode_idx": 17, + "seed": 59, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 18, + "task_name": "CheesyBread", + "episode_idx": 18, + "seed": 60, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 25.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 19, + "task_name": "CheesyBread", + "episode_idx": 19, + "seed": 61, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 18.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 20, + "task_name": "CheesyBread", + "episode_idx": 20, + "seed": 62, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 21, + "task_name": "CheesyBread", + "episode_idx": 21, + "seed": 63, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 22, + "task_name": "CheesyBread", + "episode_idx": 22, + "seed": 64, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 20.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 23, + "task_name": "CheesyBread", + "episode_idx": 23, + "seed": 65, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 10.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 24, + "task_name": "CheesyBread", + "episode_idx": 24, + "seed": 66, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 25, + "task_name": "CheesyBread", + "episode_idx": 25, + "seed": 67, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 26, + "task_name": "CheesyBread", + "episode_idx": 26, + "seed": 68, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 27, + "task_name": "CheesyBread", + "episode_idx": 27, + "seed": 69, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 28, + "task_name": "CheesyBread", + "episode_idx": 28, + "seed": 70, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 29, + "task_name": "CheesyBread", + "episode_idx": 29, + "seed": 71, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 30, + "task_name": "CheesyBread", + "episode_idx": 30, + "seed": 72, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 24.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 31, + "task_name": "CheesyBread", + "episode_idx": 31, + "seed": 73, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 32, + "task_name": "CheesyBread", + "episode_idx": 32, + "seed": 74, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 28.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 33, + "task_name": "CheesyBread", + "episode_idx": 33, + "seed": 75, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 24.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 34, + "task_name": "CheesyBread", + "episode_idx": 34, + "seed": 76, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 35, + "task_name": "CheesyBread", + "episode_idx": 35, + "seed": 77, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 36, + "task_name": "CheesyBread", + "episode_idx": 36, + "seed": 78, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 37, + "task_name": "CheesyBread", + "episode_idx": 37, + "seed": 79, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 29.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 38, + "task_name": "CheesyBread", + "episode_idx": 38, + "seed": 80, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 39, + "task_name": "CheesyBread", + "episode_idx": 39, + "seed": 81, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 28.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 40, + "task_name": "CheesyBread", + "episode_idx": 40, + "seed": 82, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 41, + "task_name": "CheesyBread", + "episode_idx": 41, + "seed": 83, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 31.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 42, + "task_name": "CheesyBread", + "episode_idx": 42, + "seed": 84, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 43, + "task_name": "CheesyBread", + "episode_idx": 43, + "seed": 85, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 28.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 44, + "task_name": "CheesyBread", + "episode_idx": 44, + "seed": 86, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 45, + "task_name": "CheesyBread", + "episode_idx": 45, + "seed": 87, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 46, + "task_name": "CheesyBread", + "episode_idx": 46, + "seed": 88, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 21.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 47, + "task_name": "CheesyBread", + "episode_idx": 47, + "seed": 89, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 48, + "task_name": "CheesyBread", + "episode_idx": 48, + "seed": 90, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 21.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 49, + "task_name": "CheesyBread", + "episode_idx": 49, + "seed": 91, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 9, + "video_path": null + } +] \ No newline at end of file diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/CheesyBread/stats.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/CheesyBread/stats.json new file mode 100644 index 0000000000000000000000000000000000000000..e02bfae8dc9ea0a85fb9ff028e7d82576a9a17f5 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/CheesyBread/stats.json @@ -0,0 +1,606 @@ +{ + "num_episodes": 50, + "success_rate": 0.4, + "episodes": [ + { + "completion_order": 0, + "task_name": "CheesyBread", + "episode_idx": 0, + "seed": 42, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 1, + "task_name": "CheesyBread", + "episode_idx": 1, + "seed": 43, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 2, + "task_name": "CheesyBread", + "episode_idx": 2, + "seed": 44, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 27.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 3, + "task_name": "CheesyBread", + "episode_idx": 3, + "seed": 45, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 31.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 4, + "task_name": "CheesyBread", + "episode_idx": 4, + "seed": 46, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 5, + "task_name": "CheesyBread", + "episode_idx": 5, + "seed": 47, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 27.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 6, + "task_name": "CheesyBread", + "episode_idx": 6, + "seed": 48, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 26.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 7, + "task_name": "CheesyBread", + "episode_idx": 7, + "seed": 49, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 8, + "task_name": "CheesyBread", + "episode_idx": 8, + "seed": 50, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 9, + "task_name": "CheesyBread", + "episode_idx": 9, + "seed": 51, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 29.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 10, + "task_name": "CheesyBread", + "episode_idx": 10, + "seed": 52, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 11, + "task_name": "CheesyBread", + "episode_idx": 11, + "seed": 53, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 12, + "task_name": "CheesyBread", + "episode_idx": 12, + "seed": 54, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 13, + "task_name": "CheesyBread", + "episode_idx": 13, + "seed": 55, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 14, + "task_name": "CheesyBread", + "episode_idx": 14, + "seed": 56, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 29.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 15, + "task_name": "CheesyBread", + "episode_idx": 15, + "seed": 57, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 32.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 16, + "task_name": "CheesyBread", + "episode_idx": 16, + "seed": 58, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 17, + "task_name": "CheesyBread", + "episode_idx": 17, + "seed": 59, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 18, + "task_name": "CheesyBread", + "episode_idx": 18, + "seed": 60, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 25.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 19, + "task_name": "CheesyBread", + "episode_idx": 19, + "seed": 61, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 18.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 20, + "task_name": "CheesyBread", + "episode_idx": 20, + "seed": 62, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 21, + "task_name": "CheesyBread", + "episode_idx": 21, + "seed": 63, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 22, + "task_name": "CheesyBread", + "episode_idx": 22, + "seed": 64, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 20.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 23, + "task_name": "CheesyBread", + "episode_idx": 23, + "seed": 65, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 10.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 24, + "task_name": "CheesyBread", + "episode_idx": 24, + "seed": 66, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 25, + "task_name": "CheesyBread", + "episode_idx": 25, + "seed": 67, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 26, + "task_name": "CheesyBread", + "episode_idx": 26, + "seed": 68, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 27, + "task_name": "CheesyBread", + "episode_idx": 27, + "seed": 69, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 28, + "task_name": "CheesyBread", + "episode_idx": 28, + "seed": 70, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 29, + "task_name": "CheesyBread", + "episode_idx": 29, + "seed": 71, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 30, + "task_name": "CheesyBread", + "episode_idx": 30, + "seed": 72, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 24.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 31, + "task_name": "CheesyBread", + "episode_idx": 31, + "seed": 73, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 32, + "task_name": "CheesyBread", + "episode_idx": 32, + "seed": 74, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 28.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 33, + "task_name": "CheesyBread", + "episode_idx": 33, + "seed": 75, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 24.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 34, + "task_name": "CheesyBread", + "episode_idx": 34, + "seed": 76, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 35, + "task_name": "CheesyBread", + "episode_idx": 35, + "seed": 77, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 36, + "task_name": "CheesyBread", + "episode_idx": 36, + "seed": 78, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 37, + "task_name": "CheesyBread", + "episode_idx": 37, + "seed": 79, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 29.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 38, + "task_name": "CheesyBread", + "episode_idx": 38, + "seed": 80, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 39, + "task_name": "CheesyBread", + "episode_idx": 39, + "seed": 81, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 28.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 40, + "task_name": "CheesyBread", + "episode_idx": 40, + "seed": 82, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 41, + "task_name": "CheesyBread", + "episode_idx": 41, + "seed": 83, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 31.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 42, + "task_name": "CheesyBread", + "episode_idx": 42, + "seed": 84, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 43, + "task_name": "CheesyBread", + "episode_idx": 43, + "seed": 85, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 28.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 44, + "task_name": "CheesyBread", + "episode_idx": 44, + "seed": 86, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 45, + "task_name": "CheesyBread", + "episode_idx": 45, + "seed": 87, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 46, + "task_name": "CheesyBread", + "episode_idx": 46, + "seed": 88, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 21.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 47, + "task_name": "CheesyBread", + "episode_idx": 47, + "seed": 89, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 48, + "task_name": "CheesyBread", + "episode_idx": 48, + "seed": 90, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 21.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 49, + "task_name": "CheesyBread", + "episode_idx": 49, + "seed": 91, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 9, + "video_path": null + } + ] +} \ No newline at end of file diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/2cc7ab06-895a-49b2-905c-0131f40bb57a.mp4 b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/2cc7ab06-895a-49b2-905c-0131f40bb57a.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..17e5e5fbd204f4f1c8bf240b166ab0a318db4744 Binary files /dev/null and b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/2cc7ab06-895a-49b2-905c-0131f40bb57a.mp4 differ diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/44a78811-81d1-4f26-af76-9a1ba4827178.mp4 b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/44a78811-81d1-4f26-af76-9a1ba4827178.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..17e5e5fbd204f4f1c8bf240b166ab0a318db4744 Binary files /dev/null and b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/44a78811-81d1-4f26-af76-9a1ba4827178.mp4 differ diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/7583c098-2567-42a0-b01d-b73d29454f3e.mp4 b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/7583c098-2567-42a0-b01d-b73d29454f3e.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..17e5e5fbd204f4f1c8bf240b166ab0a318db4744 Binary files /dev/null and b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/7583c098-2567-42a0-b01d-b73d29454f3e.mp4 differ diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/7f2aca5c-e3b6-4e39-9c3e-2d348a011024.mp4 b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/7f2aca5c-e3b6-4e39-9c3e-2d348a011024.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..17e5e5fbd204f4f1c8bf240b166ab0a318db4744 Binary files /dev/null and b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/7f2aca5c-e3b6-4e39-9c3e-2d348a011024.mp4 differ diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/8032f648-4a07-4861-a411-fec34ac821a9.mp4 b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/8032f648-4a07-4861-a411-fec34ac821a9.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..17e5e5fbd204f4f1c8bf240b166ab0a318db4744 Binary files /dev/null and b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/8032f648-4a07-4861-a411-fec34ac821a9.mp4 differ diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/80715ec3-e819-4f8f-9924-51c9ba244bef.mp4 b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/80715ec3-e819-4f8f-9924-51c9ba244bef.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..17e5e5fbd204f4f1c8bf240b166ab0a318db4744 Binary files /dev/null and b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/80715ec3-e819-4f8f-9924-51c9ba244bef.mp4 differ diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/8b650368-ba33-479c-bc96-c2fc6d1ce8c7.mp4 b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/8b650368-ba33-479c-bc96-c2fc6d1ce8c7.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..17e5e5fbd204f4f1c8bf240b166ab0a318db4744 Binary files /dev/null and b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/8b650368-ba33-479c-bc96-c2fc6d1ce8c7.mp4 differ diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/ae4c66c6-da66-4338-8f21-320ef0e3a180.mp4 b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/ae4c66c6-da66-4338-8f21-320ef0e3a180.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..17e5e5fbd204f4f1c8bf240b166ab0a318db4744 Binary files /dev/null and b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/ae4c66c6-da66-4338-8f21-320ef0e3a180.mp4 differ diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/e5860fa6-917a-49d4-9926-f701d6fe072e.mp4 b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/e5860fa6-917a-49d4-9926-f701d6fe072e.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..17e5e5fbd204f4f1c8bf240b166ab0a318db4744 Binary files /dev/null and b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/PackDessert/e5860fa6-917a-49d4-9926-f701d6fe072e.mp4 differ diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/experiment_cfg/metadata.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/experiment_cfg/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..27700cac5d292d4d42897b3c2e80bdc58e55c60e --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/experiment_cfg/metadata.json @@ -0,0 +1,431 @@ +{ + "new_embodiment": { + "statistics": { + "state": { + "base_position": { + "max": [ + 7.778976917266846, + 0.33403685688972473, + 0.7223060727119446 + ], + "min": [ + -4.821934223175049, + -6.785110950469971, + 0.6994727253913879 + ], + "mean": [ + 2.5719731235192254, + -2.1489371002267235, + 0.7003612741268546 + ], + "std": [ + 1.669945884010784, + 1.3739411381361726, + 0.0009413282367754987 + ], + "q01": [ + -0.7944286339399775, + -5.486611518471242, + 0.699977555971739 + ], + "q99": [ + 6.019886905263648, + -0.5229184210026855, + 0.702930326282967 + ] + }, + "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, + 0.248543484335273, + 0.5959317405459531 + ], + "std": [ + 0.0, + 0.0, + 0.6655550240074092, + 0.374283995143559 + ], + "q01": [ + 0.0, + 0.0, + -1.0000000000000002, + 4.503254968910557e-06 + ], + "q99": [ + 0.0, + 0.0, + 1.0000000000000002, + 1.0000000000000002 + ] + }, + "end_effector_position_relative": { + "max": [ + 0.8848381638526917, + 0.8569459915161133, + 1.053884744644165 + ], + "min": [ + -0.5828549861907959, + -0.8486236333847046, + -0.3446122705936432 + ], + "mean": [ + 0.2465490874130862, + -0.004711658908481467, + 0.43463835464312073 + ], + "std": [ + 0.15798784670671595, + 0.25624102400497073, + 0.20836028727974915 + ], + "q01": [ + -0.020752051134640608, + -0.4837933887347875, + 0.1802942952289915 + ], + "q99": [ + 0.5386784714844007, + 0.4801424212024721, + 0.695095732975766 + ] + }, + "end_effector_rotation_relative": { + "max": [ + 0.9999949932098389, + 0.9997354745864868, + 0.9900747537612915, + 0.9735639095306396 + ], + "min": [ + -0.9999914169311523, + -0.9999884366989136, + -0.9975369572639465, + 6.754255821306288e-08 + ], + "mean": [ + -0.22531859714056926, + 0.046973702821026614, + -0.0568176967878771, + 0.12364307042955697 + ], + "std": [ + 0.8502144333612052, + 0.29591250318147405, + 0.30218053156837016, + 0.1636031353412931 + ], + "q01": [ + -0.9959057387172566, + -0.5665384745573048, + -0.5329334620834977, + 0.0014372279999397553 + ], + "q99": [ + 0.9904107951565839, + 0.6359623612021396, + 0.5026076022439377, + 0.42955559234879914 + ] + }, + "gripper_qpos": { + "max": [ + 0.05289580300450325, + 0.00937709677964449 + ], + "min": [ + -0.011669117026031017, + -0.06307756155729294 + ], + "mean": [ + 0.03188757060429838, + -0.0321793333086053 + ], + "std": [ + 0.01171052313737148, + 0.011427336801203189 + ], + "q01": [ + 0.006462901319459925, + -0.040557140599419166 + ], + "q99": [ + 0.04037619335388386, + -0.006523947835993347 + ] + } + }, + "action": { + "base_motion": { + "max": [ + 1.0, + 1.0, + 1.0, + 0.0 + ], + "min": [ + -1.0, + -1.0, + -1.0, + 0.0 + ], + "mean": [ + 0.006639007355857881, + 0.0027330421795561874, + -0.0017774441970214892, + 0.0 + ], + "std": [ + 0.11841528114040736, + 0.1041359083946544, + 0.08346265521208351, + 0.0 + ], + "q01": [ + -0.17562359675270237, + -0.21706402321529558, + -0.16273392585723104, + 0.0 + ], + "q99": [ + 0.2572918701644887, + 0.27022325575787115, + 0.1369626751049278, + 0.0 + ] + }, + "control_mode": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.9205539608927732 + ], + "std": [ + 0.39063657242534666 + ], + "q01": [ + -1.0 + ], + "q99": [ + -0.17711863866562935 + ] + }, + "end_effector_position": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + -0.00924803964621179, + 0.004053841761390575, + -0.08385115578460103 + ], + "std": [ + 0.45413072518223824, + 0.40158612714634856, + 0.41068634292849304 + ], + "q01": [ + -0.9669817145875343, + -0.9241307799783718, + -0.8264079503068825 + ], + "q99": [ + 0.8021511975722166, + 0.9241346582748478, + 0.8348116781502298 + ] + }, + "end_effector_rotation": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.007644157504353809, + -0.019962543093886335, + -0.002047110452602617 + ], + "std": [ + 0.11076012154791087, + 0.13176735140285795, + 0.12286528665895614 + ], + "q01": [ + -0.2659105439401192, + -0.3961694033813533, + -0.3434870832468936 + ], + "q99": [ + 0.3145786959266369, + 0.31309430361227547, + 0.34107754401942986 + ] + }, + "gripper_close": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.29870612267533575 + ], + "std": [ + 0.9543487555375691 + ], + "q01": [ + -1.0 + ], + "q99": [ + 0.7610674802484976 + ] + } + } + }, + "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/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/model.safetensors.index.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..5ef5f247b61eb93e89601c7da8334fc087e19750 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/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/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/trainer_state.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..77114f6884a322afd5643f3643a6949d1e5a2034 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed1/default/train_5shot_seed1_processingline_12k/trainer_state.json @@ -0,0 +1,8443 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 20.100502512562816, + "eval_steps": 500, + "global_step": 12000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.01675041876046901, + "grad_norm": 1.1124801635742188, + "learning_rate": 4.5e-07, + "loss": 0.2576, + "step": 10 + }, + { + "epoch": 0.03350083752093802, + "grad_norm": 1.3764758110046387, + "learning_rate": 9.500000000000001e-07, + "loss": 0.2541, + "step": 20 + }, + { + "epoch": 0.05025125628140704, + "grad_norm": 1.0254155397415161, + "learning_rate": 1.45e-06, + "loss": 0.2369, + "step": 30 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 0.6008508801460266, + "learning_rate": 1.95e-06, + "loss": 0.2272, + "step": 40 + }, + { + "epoch": 0.08375209380234507, + "grad_norm": 1.085799217224121, + "learning_rate": 2.45e-06, + "loss": 0.2211, + "step": 50 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 0.9870124459266663, + "learning_rate": 2.9499999999999997e-06, + "loss": 0.2511, + "step": 60 + }, + { + "epoch": 0.11725293132328309, + "grad_norm": 0.9730237722396851, + "learning_rate": 3.4500000000000004e-06, + "loss": 0.2414, + "step": 70 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 0.792273998260498, + "learning_rate": 3.9499999999999995e-06, + "loss": 0.2311, + "step": 80 + }, + { + "epoch": 0.1507537688442211, + "grad_norm": 1.078325629234314, + "learning_rate": 4.450000000000001e-06, + "loss": 0.2185, + "step": 90 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 0.6237806081771851, + "learning_rate": 4.95e-06, + "loss": 0.1754, + "step": 100 + }, + { + "epoch": 0.18425460636515914, + "grad_norm": 0.7615986466407776, + "learning_rate": 5.45e-06, + "loss": 0.167, + "step": 110 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 0.6572969555854797, + "learning_rate": 5.95e-06, + "loss": 0.1568, + "step": 120 + }, + { + "epoch": 0.21775544388609716, + "grad_norm": 0.5472678542137146, + "learning_rate": 6.45e-06, + "loss": 0.1421, + "step": 130 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 0.4665081202983856, + "learning_rate": 6.95e-06, + "loss": 0.1526, + "step": 140 + }, + { + "epoch": 0.25125628140703515, + "grad_norm": 0.5054625272750854, + "learning_rate": 7.45e-06, + "loss": 0.1246, + "step": 150 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 0.5371837019920349, + "learning_rate": 7.95e-06, + "loss": 0.1179, + "step": 160 + }, + { + "epoch": 0.2847571189279732, + "grad_norm": 0.40661293268203735, + "learning_rate": 8.45e-06, + "loss": 0.1062, + "step": 170 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 0.2422139048576355, + "learning_rate": 8.95e-06, + "loss": 0.0963, + "step": 180 + }, + { + "epoch": 0.31825795644891125, + "grad_norm": 0.3034665584564209, + "learning_rate": 9.450000000000001e-06, + "loss": 0.0975, + "step": 190 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 0.2536943256855011, + "learning_rate": 9.95e-06, + "loss": 0.0934, + "step": 200 + }, + { + "epoch": 0.35175879396984927, + "grad_norm": 0.2668403685092926, + "learning_rate": 1.045e-05, + "loss": 0.0933, + "step": 210 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 0.22533497214317322, + "learning_rate": 1.095e-05, + "loss": 0.0792, + "step": 220 + }, + { + "epoch": 0.3852596314907873, + "grad_norm": 0.2040107697248459, + "learning_rate": 1.145e-05, + "loss": 0.0812, + "step": 230 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 0.19936227798461914, + "learning_rate": 1.1949999999999999e-05, + "loss": 0.0838, + "step": 240 + }, + { + "epoch": 0.4187604690117253, + "grad_norm": 0.2014681100845337, + "learning_rate": 1.245e-05, + "loss": 0.0804, + "step": 250 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 0.20712223649024963, + "learning_rate": 1.295e-05, + "loss": 0.0751, + "step": 260 + }, + { + "epoch": 0.45226130653266333, + "grad_norm": 0.23043803870677948, + "learning_rate": 1.345e-05, + "loss": 0.0747, + "step": 270 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 0.19819584488868713, + "learning_rate": 1.395e-05, + "loss": 0.0715, + "step": 280 + }, + { + "epoch": 0.48576214405360135, + "grad_norm": 0.2228541076183319, + "learning_rate": 1.445e-05, + "loss": 0.0708, + "step": 290 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 0.17452232539653778, + "learning_rate": 1.4950000000000001e-05, + "loss": 0.0685, + "step": 300 + }, + { + "epoch": 0.5192629815745393, + "grad_norm": 0.20126119256019592, + "learning_rate": 1.545e-05, + "loss": 0.0703, + "step": 310 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 0.2398366630077362, + "learning_rate": 1.5949999999999998e-05, + "loss": 0.0664, + "step": 320 + }, + { + "epoch": 0.5527638190954773, + "grad_norm": 0.21752075850963593, + "learning_rate": 1.645e-05, + "loss": 0.0679, + "step": 330 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 0.17013588547706604, + "learning_rate": 1.695e-05, + "loss": 0.0705, + "step": 340 + }, + { + "epoch": 0.5862646566164154, + "grad_norm": 0.2016950249671936, + "learning_rate": 1.745e-05, + "loss": 0.0687, + "step": 350 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 0.1707467883825302, + "learning_rate": 1.7950000000000003e-05, + "loss": 0.0675, + "step": 360 + }, + { + "epoch": 0.6197654941373534, + "grad_norm": 0.17180311679840088, + "learning_rate": 1.845e-05, + "loss": 0.0618, + "step": 370 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 0.21700109541416168, + "learning_rate": 1.8950000000000003e-05, + "loss": 0.0631, + "step": 380 + }, + { + "epoch": 0.6532663316582915, + "grad_norm": 0.21110156178474426, + "learning_rate": 1.945e-05, + "loss": 0.061, + "step": 390 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 0.17809593677520752, + "learning_rate": 1.995e-05, + "loss": 0.0585, + "step": 400 + }, + { + "epoch": 0.6867671691792295, + "grad_norm": 0.15697689354419708, + "learning_rate": 2.045e-05, + "loss": 0.0615, + "step": 410 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 0.17097905278205872, + "learning_rate": 2.095e-05, + "loss": 0.0568, + "step": 420 + }, + { + "epoch": 0.7202680067001676, + "grad_norm": 0.17003680765628815, + "learning_rate": 2.145e-05, + "loss": 0.0609, + "step": 430 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 0.20018595457077026, + "learning_rate": 2.195e-05, + "loss": 0.0538, + "step": 440 + }, + { + "epoch": 0.7537688442211056, + "grad_norm": 0.1872512400150299, + "learning_rate": 2.245e-05, + "loss": 0.0562, + "step": 450 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 0.210875004529953, + "learning_rate": 2.2950000000000002e-05, + "loss": 0.0538, + "step": 460 + }, + { + "epoch": 0.7872696817420436, + "grad_norm": 0.19456243515014648, + "learning_rate": 2.345e-05, + "loss": 0.0548, + "step": 470 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 0.19903069734573364, + "learning_rate": 2.395e-05, + "loss": 0.0536, + "step": 480 + }, + { + "epoch": 0.8207705192629816, + "grad_norm": 0.1998818963766098, + "learning_rate": 2.4449999999999998e-05, + "loss": 0.0533, + "step": 490 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 0.22372868657112122, + "learning_rate": 2.495e-05, + "loss": 0.0532, + "step": 500 + }, + { + "epoch": 0.8542713567839196, + "grad_norm": 0.2313748300075531, + "learning_rate": 2.5450000000000002e-05, + "loss": 0.049, + "step": 510 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 0.1595960259437561, + "learning_rate": 2.595e-05, + "loss": 0.0513, + "step": 520 + }, + { + "epoch": 0.8877721943048577, + "grad_norm": 0.3054298162460327, + "learning_rate": 2.6450000000000003e-05, + "loss": 0.0489, + "step": 530 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 0.15157485008239746, + "learning_rate": 2.695e-05, + "loss": 0.0501, + "step": 540 + }, + { + "epoch": 0.9212730318257957, + "grad_norm": 0.19291409850120544, + "learning_rate": 2.7450000000000003e-05, + "loss": 0.0428, + "step": 550 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 0.22616443037986755, + "learning_rate": 2.795e-05, + "loss": 0.048, + "step": 560 + }, + { + "epoch": 0.9547738693467337, + "grad_norm": 0.21530359983444214, + "learning_rate": 2.845e-05, + "loss": 0.051, + "step": 570 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 0.3027316629886627, + "learning_rate": 2.895e-05, + "loss": 0.0462, + "step": 580 + }, + { + "epoch": 0.9882747068676717, + "grad_norm": 0.16115690767765045, + "learning_rate": 2.945e-05, + "loss": 0.0438, + "step": 590 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 0.20157869160175323, + "learning_rate": 2.995e-05, + "loss": 0.0451, + "step": 600 + }, + { + "epoch": 1.0217755443886096, + "grad_norm": 0.19205620884895325, + "learning_rate": 2.9999953864407506e-05, + "loss": 0.0437, + "step": 610 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 0.1747637391090393, + "learning_rate": 2.99997943837114e-05, + "loss": 0.0454, + "step": 620 + }, + { + "epoch": 1.0552763819095476, + "grad_norm": 0.18132591247558594, + "learning_rate": 2.9999520989547326e-05, + "loss": 0.0488, + "step": 630 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 0.1378272920846939, + "learning_rate": 2.9999133683991525e-05, + "loss": 0.042, + "step": 640 + }, + { + "epoch": 1.0887772194304857, + "grad_norm": 0.17927886545658112, + "learning_rate": 2.9998632469985326e-05, + "loss": 0.0452, + "step": 650 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 0.1842069774866104, + "learning_rate": 2.999801735133512e-05, + "loss": 0.0418, + "step": 660 + }, + { + "epoch": 1.1222780569514237, + "grad_norm": 0.1894877851009369, + "learning_rate": 2.9997288332712325e-05, + "loss": 0.0421, + "step": 670 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 0.17502732574939728, + "learning_rate": 2.999644541965335e-05, + "loss": 0.0395, + "step": 680 + }, + { + "epoch": 1.1557788944723617, + "grad_norm": 0.19280853867530823, + "learning_rate": 2.9995488618559554e-05, + "loss": 0.0383, + "step": 690 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 0.21065080165863037, + "learning_rate": 2.9994417936697218e-05, + "loss": 0.0409, + "step": 700 + }, + { + "epoch": 1.1892797319932997, + "grad_norm": 0.15386411547660828, + "learning_rate": 2.9993233382197454e-05, + "loss": 0.0417, + "step": 710 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 0.18249723315238953, + "learning_rate": 2.999193496405616e-05, + "loss": 0.0386, + "step": 720 + }, + { + "epoch": 1.2227805695142377, + "grad_norm": 0.17197810113430023, + "learning_rate": 2.9990522692133962e-05, + "loss": 0.038, + "step": 730 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 0.15405157208442688, + "learning_rate": 2.9988996577156134e-05, + "loss": 0.0421, + "step": 740 + }, + { + "epoch": 1.2562814070351758, + "grad_norm": 0.16718165576457977, + "learning_rate": 2.9987356630712496e-05, + "loss": 0.04, + "step": 750 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 0.1574426144361496, + "learning_rate": 2.9985602865257354e-05, + "loss": 0.0403, + "step": 760 + }, + { + "epoch": 1.2897822445561138, + "grad_norm": 0.11435810476541519, + "learning_rate": 2.998373529410939e-05, + "loss": 0.0398, + "step": 770 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 0.16009581089019775, + "learning_rate": 2.9981753931451567e-05, + "loss": 0.037, + "step": 780 + }, + { + "epoch": 1.3232830820770518, + "grad_norm": 0.15292680263519287, + "learning_rate": 2.9979658792331016e-05, + "loss": 0.0375, + "step": 790 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 0.1592884510755539, + "learning_rate": 2.997744989265893e-05, + "loss": 0.037, + "step": 800 + }, + { + "epoch": 1.3567839195979898, + "grad_norm": 0.22414714097976685, + "learning_rate": 2.9975127249210426e-05, + "loss": 0.0344, + "step": 810 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 0.14030888676643372, + "learning_rate": 2.9972690879624445e-05, + "loss": 0.0364, + "step": 820 + }, + { + "epoch": 1.3902847571189278, + "grad_norm": 0.15426629781723022, + "learning_rate": 2.9970140802403587e-05, + "loss": 0.0368, + "step": 830 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 0.1377844214439392, + "learning_rate": 2.9967477036914002e-05, + "loss": 0.0365, + "step": 840 + }, + { + "epoch": 1.4237855946398659, + "grad_norm": 0.1314632147550583, + "learning_rate": 2.996469960338521e-05, + "loss": 0.0364, + "step": 850 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 0.1743793785572052, + "learning_rate": 2.9961808522909976e-05, + "loss": 0.0383, + "step": 860 + }, + { + "epoch": 1.457286432160804, + "grad_norm": 0.1816023737192154, + "learning_rate": 2.995880381744414e-05, + "loss": 0.0357, + "step": 870 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 0.19835101068019867, + "learning_rate": 2.995568550980643e-05, + "loss": 0.0383, + "step": 880 + }, + { + "epoch": 1.490787269681742, + "grad_norm": 0.13531947135925293, + "learning_rate": 2.9952453623678334e-05, + "loss": 0.0338, + "step": 890 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 0.13880357146263123, + "learning_rate": 2.9949108183603867e-05, + "loss": 0.036, + "step": 900 + }, + { + "epoch": 1.52428810720268, + "grad_norm": 0.18743862211704254, + "learning_rate": 2.994564921498943e-05, + "loss": 0.0358, + "step": 910 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 0.1355157345533371, + "learning_rate": 2.9942076744103594e-05, + "loss": 0.0342, + "step": 920 + }, + { + "epoch": 1.557788944723618, + "grad_norm": 0.16391365230083466, + "learning_rate": 2.9938390798076886e-05, + "loss": 0.0359, + "step": 930 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 0.1717372089624405, + "learning_rate": 2.993459140490163e-05, + "loss": 0.0344, + "step": 940 + }, + { + "epoch": 1.591289782244556, + "grad_norm": 0.19507038593292236, + "learning_rate": 2.9930678593431684e-05, + "loss": 0.0325, + "step": 950 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 0.14840291440486908, + "learning_rate": 2.9926652393382258e-05, + "loss": 0.0334, + "step": 960 + }, + { + "epoch": 1.624790619765494, + "grad_norm": 0.15482094883918762, + "learning_rate": 2.9922512835329655e-05, + "loss": 0.0338, + "step": 970 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 0.12408526986837387, + "learning_rate": 2.9918259950711073e-05, + "loss": 0.0324, + "step": 980 + }, + { + "epoch": 1.658291457286432, + "grad_norm": 0.1402246356010437, + "learning_rate": 2.991389377182435e-05, + "loss": 0.0363, + "step": 990 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 0.15740147233009338, + "learning_rate": 2.99094143318277e-05, + "loss": 0.0328, + "step": 1000 + }, + { + "epoch": 1.69179229480737, + "grad_norm": 0.14837925136089325, + "learning_rate": 2.9904821664739493e-05, + "loss": 0.0332, + "step": 1010 + }, + { + "epoch": 1.708542713567839, + "grad_norm": 0.1531188040971756, + "learning_rate": 2.990011580543799e-05, + "loss": 0.0326, + "step": 1020 + }, + { + "epoch": 1.725293132328308, + "grad_norm": 0.1466841697692871, + "learning_rate": 2.9895296789661065e-05, + "loss": 0.0326, + "step": 1030 + }, + { + "epoch": 1.742043551088777, + "grad_norm": 0.14380747079849243, + "learning_rate": 2.989036465400594e-05, + "loss": 0.0311, + "step": 1040 + }, + { + "epoch": 1.758793969849246, + "grad_norm": 0.15117715299129486, + "learning_rate": 2.9885319435928896e-05, + "loss": 0.0308, + "step": 1050 + }, + { + "epoch": 1.775544388609715, + "grad_norm": 0.1409190148115158, + "learning_rate": 2.9880161173745017e-05, + "loss": 0.0288, + "step": 1060 + }, + { + "epoch": 1.792294807370184, + "grad_norm": 0.11704489588737488, + "learning_rate": 2.9874889906627882e-05, + "loss": 0.0314, + "step": 1070 + }, + { + "epoch": 1.809045226130653, + "grad_norm": 0.1579095870256424, + "learning_rate": 2.9869505674609253e-05, + "loss": 0.0334, + "step": 1080 + }, + { + "epoch": 1.8257956448911221, + "grad_norm": 0.14541688561439514, + "learning_rate": 2.9864008518578794e-05, + "loss": 0.0301, + "step": 1090 + }, + { + "epoch": 1.8425460636515911, + "grad_norm": 0.11781132966279984, + "learning_rate": 2.9858398480283758e-05, + "loss": 0.031, + "step": 1100 + }, + { + "epoch": 1.8592964824120601, + "grad_norm": 0.1376485973596573, + "learning_rate": 2.9852675602328644e-05, + "loss": 0.0296, + "step": 1110 + }, + { + "epoch": 1.8760469011725294, + "grad_norm": 0.12675292789936066, + "learning_rate": 2.9846839928174923e-05, + "loss": 0.0289, + "step": 1120 + }, + { + "epoch": 1.8927973199329984, + "grad_norm": 0.14450091123580933, + "learning_rate": 2.984089150214065e-05, + "loss": 0.0297, + "step": 1130 + }, + { + "epoch": 1.9095477386934674, + "grad_norm": 0.15321670472621918, + "learning_rate": 2.9834830369400165e-05, + "loss": 0.0295, + "step": 1140 + }, + { + "epoch": 1.9262981574539364, + "grad_norm": 0.13381171226501465, + "learning_rate": 2.9828656575983747e-05, + "loss": 0.0279, + "step": 1150 + }, + { + "epoch": 1.9430485762144054, + "grad_norm": 0.20773489773273468, + "learning_rate": 2.9822370168777253e-05, + "loss": 0.0305, + "step": 1160 + }, + { + "epoch": 1.9597989949748744, + "grad_norm": 0.1649148315191269, + "learning_rate": 2.981597119552177e-05, + "loss": 0.0318, + "step": 1170 + }, + { + "epoch": 1.9765494137353434, + "grad_norm": 0.15318723022937775, + "learning_rate": 2.9809459704813236e-05, + "loss": 0.0283, + "step": 1180 + }, + { + "epoch": 1.9932998324958124, + "grad_norm": 0.11235921084880829, + "learning_rate": 2.980283574610211e-05, + "loss": 0.0281, + "step": 1190 + }, + { + "epoch": 2.0100502512562812, + "grad_norm": 0.14214928448200226, + "learning_rate": 2.9796099369692946e-05, + "loss": 0.0277, + "step": 1200 + }, + { + "epoch": 2.0268006700167502, + "grad_norm": 0.1358565390110016, + "learning_rate": 2.978925062674404e-05, + "loss": 0.0281, + "step": 1210 + }, + { + "epoch": 2.0435510887772192, + "grad_norm": 0.1451619565486908, + "learning_rate": 2.9782289569267057e-05, + "loss": 0.0279, + "step": 1220 + }, + { + "epoch": 2.0603015075376883, + "grad_norm": 0.1291060596704483, + "learning_rate": 2.977521625012659e-05, + "loss": 0.0307, + "step": 1230 + }, + { + "epoch": 2.0770519262981573, + "grad_norm": 0.13382935523986816, + "learning_rate": 2.9768030723039798e-05, + "loss": 0.0281, + "step": 1240 + }, + { + "epoch": 2.0938023450586263, + "grad_norm": 0.1389288753271103, + "learning_rate": 2.9760733042575984e-05, + "loss": 0.0286, + "step": 1250 + }, + { + "epoch": 2.1105527638190953, + "grad_norm": 0.13549239933490753, + "learning_rate": 2.9753323264156184e-05, + "loss": 0.0283, + "step": 1260 + }, + { + "epoch": 2.1273031825795643, + "grad_norm": 0.13530868291854858, + "learning_rate": 2.974580144405273e-05, + "loss": 0.0281, + "step": 1270 + }, + { + "epoch": 2.1440536013400333, + "grad_norm": 0.12069135159254074, + "learning_rate": 2.973816763938886e-05, + "loss": 0.0308, + "step": 1280 + }, + { + "epoch": 2.1608040201005023, + "grad_norm": 0.15732339024543762, + "learning_rate": 2.9730421908138236e-05, + "loss": 0.0274, + "step": 1290 + }, + { + "epoch": 2.1775544388609713, + "grad_norm": 0.14720749855041504, + "learning_rate": 2.9722564309124546e-05, + "loss": 0.0284, + "step": 1300 + }, + { + "epoch": 2.1943048576214403, + "grad_norm": 0.13191711902618408, + "learning_rate": 2.971459490202102e-05, + "loss": 0.0259, + "step": 1310 + }, + { + "epoch": 2.2110552763819094, + "grad_norm": 0.12703394889831543, + "learning_rate": 2.9706513747350017e-05, + "loss": 0.0245, + "step": 1320 + }, + { + "epoch": 2.2278056951423784, + "grad_norm": 0.13217298686504364, + "learning_rate": 2.9698320906482535e-05, + "loss": 0.0307, + "step": 1330 + }, + { + "epoch": 2.2445561139028474, + "grad_norm": 0.12877672910690308, + "learning_rate": 2.9690016441637747e-05, + "loss": 0.0276, + "step": 1340 + }, + { + "epoch": 2.2613065326633164, + "grad_norm": 0.13202929496765137, + "learning_rate": 2.9681600415882556e-05, + "loss": 0.0289, + "step": 1350 + }, + { + "epoch": 2.2780569514237854, + "grad_norm": 0.11203215271234512, + "learning_rate": 2.9673072893131068e-05, + "loss": 0.0274, + "step": 1360 + }, + { + "epoch": 2.2948073701842544, + "grad_norm": 0.09884139150381088, + "learning_rate": 2.9664433938144162e-05, + "loss": 0.0275, + "step": 1370 + }, + { + "epoch": 2.3115577889447234, + "grad_norm": 0.12471596896648407, + "learning_rate": 2.9655683616528958e-05, + "loss": 0.0252, + "step": 1380 + }, + { + "epoch": 2.3283082077051924, + "grad_norm": 0.15619947016239166, + "learning_rate": 2.9646821994738322e-05, + "loss": 0.0267, + "step": 1390 + }, + { + "epoch": 2.3450586264656614, + "grad_norm": 0.11800948530435562, + "learning_rate": 2.9637849140070393e-05, + "loss": 0.025, + "step": 1400 + }, + { + "epoch": 2.3618090452261304, + "grad_norm": 0.12330026179552078, + "learning_rate": 2.9628765120668033e-05, + "loss": 0.0242, + "step": 1410 + }, + { + "epoch": 2.3785594639865995, + "grad_norm": 0.14527784287929535, + "learning_rate": 2.9619570005518336e-05, + "loss": 0.0258, + "step": 1420 + }, + { + "epoch": 2.3953098827470685, + "grad_norm": 0.1384161412715912, + "learning_rate": 2.9610263864452077e-05, + "loss": 0.0245, + "step": 1430 + }, + { + "epoch": 2.4120603015075375, + "grad_norm": 0.14128261804580688, + "learning_rate": 2.9600846768143225e-05, + "loss": 0.0243, + "step": 1440 + }, + { + "epoch": 2.4288107202680065, + "grad_norm": 0.12654323875904083, + "learning_rate": 2.959131878810836e-05, + "loss": 0.0257, + "step": 1450 + }, + { + "epoch": 2.4455611390284755, + "grad_norm": 0.19983522593975067, + "learning_rate": 2.9581679996706157e-05, + "loss": 0.0278, + "step": 1460 + }, + { + "epoch": 2.4623115577889445, + "grad_norm": 0.14568129181861877, + "learning_rate": 2.9571930467136833e-05, + "loss": 0.0256, + "step": 1470 + }, + { + "epoch": 2.4790619765494135, + "grad_norm": 0.12198968231678009, + "learning_rate": 2.956207027344159e-05, + "loss": 0.0257, + "step": 1480 + }, + { + "epoch": 2.4958123953098825, + "grad_norm": 0.11267124861478806, + "learning_rate": 2.9552099490502026e-05, + "loss": 0.0246, + "step": 1490 + }, + { + "epoch": 2.5125628140703515, + "grad_norm": 0.11769300699234009, + "learning_rate": 2.9542018194039634e-05, + "loss": 0.0249, + "step": 1500 + }, + { + "epoch": 2.5293132328308205, + "grad_norm": 0.11569931358098984, + "learning_rate": 2.953182646061514e-05, + "loss": 0.0278, + "step": 1510 + }, + { + "epoch": 2.5460636515912896, + "grad_norm": 0.12251350283622742, + "learning_rate": 2.9521524367627994e-05, + "loss": 0.0233, + "step": 1520 + }, + { + "epoch": 2.5628140703517586, + "grad_norm": 0.11296738684177399, + "learning_rate": 2.9511111993315743e-05, + "loss": 0.0251, + "step": 1530 + }, + { + "epoch": 2.5795644891122276, + "grad_norm": 0.11284096539020538, + "learning_rate": 2.9500589416753447e-05, + "loss": 0.0261, + "step": 1540 + }, + { + "epoch": 2.5963149078726966, + "grad_norm": 0.14394700527191162, + "learning_rate": 2.9489956717853088e-05, + "loss": 0.0272, + "step": 1550 + }, + { + "epoch": 2.6130653266331656, + "grad_norm": 0.12496906518936157, + "learning_rate": 2.9479213977362935e-05, + "loss": 0.0255, + "step": 1560 + }, + { + "epoch": 2.6298157453936346, + "grad_norm": 0.1484213024377823, + "learning_rate": 2.9468361276866973e-05, + "loss": 0.0257, + "step": 1570 + }, + { + "epoch": 2.6465661641541036, + "grad_norm": 0.13880600035190582, + "learning_rate": 2.9457398698784243e-05, + "loss": 0.0226, + "step": 1580 + }, + { + "epoch": 2.6633165829145726, + "grad_norm": 0.1260875016450882, + "learning_rate": 2.944632632636823e-05, + "loss": 0.026, + "step": 1590 + }, + { + "epoch": 2.6800670016750416, + "grad_norm": 0.14514032006263733, + "learning_rate": 2.9435144243706255e-05, + "loss": 0.0251, + "step": 1600 + }, + { + "epoch": 2.6968174204355106, + "grad_norm": 0.10520205646753311, + "learning_rate": 2.9423852535718805e-05, + "loss": 0.023, + "step": 1610 + }, + { + "epoch": 2.7135678391959797, + "grad_norm": 0.09249555319547653, + "learning_rate": 2.9412451288158883e-05, + "loss": 0.0238, + "step": 1620 + }, + { + "epoch": 2.7303182579564487, + "grad_norm": 0.08644948154687881, + "learning_rate": 2.94009405876114e-05, + "loss": 0.0242, + "step": 1630 + }, + { + "epoch": 2.7470686767169177, + "grad_norm": 0.12189530581235886, + "learning_rate": 2.9389320521492464e-05, + "loss": 0.0216, + "step": 1640 + }, + { + "epoch": 2.7638190954773867, + "grad_norm": 0.10813789814710617, + "learning_rate": 2.9377591178048765e-05, + "loss": 0.0249, + "step": 1650 + }, + { + "epoch": 2.7805695142378557, + "grad_norm": 0.10432349145412445, + "learning_rate": 2.9365752646356857e-05, + "loss": 0.0216, + "step": 1660 + }, + { + "epoch": 2.7973199329983247, + "grad_norm": 0.08932849764823914, + "learning_rate": 2.935380501632253e-05, + "loss": 0.0245, + "step": 1670 + }, + { + "epoch": 2.8140703517587937, + "grad_norm": 0.12938445806503296, + "learning_rate": 2.934174837868008e-05, + "loss": 0.0252, + "step": 1680 + }, + { + "epoch": 2.8308207705192627, + "grad_norm": 0.20361070334911346, + "learning_rate": 2.932958282499167e-05, + "loss": 0.0258, + "step": 1690 + }, + { + "epoch": 2.8475711892797317, + "grad_norm": 0.12765756249427795, + "learning_rate": 2.9317308447646585e-05, + "loss": 0.0256, + "step": 1700 + }, + { + "epoch": 2.8643216080402008, + "grad_norm": 0.13113832473754883, + "learning_rate": 2.9304925339860565e-05, + "loss": 0.0258, + "step": 1710 + }, + { + "epoch": 2.8810720268006698, + "grad_norm": 0.13574381172657013, + "learning_rate": 2.929243359567509e-05, + "loss": 0.0231, + "step": 1720 + }, + { + "epoch": 2.8978224455611388, + "grad_norm": 0.15551957488059998, + "learning_rate": 2.9279833309956637e-05, + "loss": 0.0233, + "step": 1730 + }, + { + "epoch": 2.914572864321608, + "grad_norm": 0.1148996576666832, + "learning_rate": 2.9267124578396013e-05, + "loss": 0.0215, + "step": 1740 + }, + { + "epoch": 2.931323283082077, + "grad_norm": 0.1132749691605568, + "learning_rate": 2.9254307497507584e-05, + "loss": 0.0255, + "step": 1750 + }, + { + "epoch": 2.948073701842546, + "grad_norm": 0.12405268102884293, + "learning_rate": 2.924138216462856e-05, + "loss": 0.0232, + "step": 1760 + }, + { + "epoch": 2.964824120603015, + "grad_norm": 0.13782398402690887, + "learning_rate": 2.922834867791825e-05, + "loss": 0.0257, + "step": 1770 + }, + { + "epoch": 2.981574539363484, + "grad_norm": 0.13731901347637177, + "learning_rate": 2.9215207136357322e-05, + "loss": 0.0235, + "step": 1780 + }, + { + "epoch": 2.998324958123953, + "grad_norm": 0.11207369714975357, + "learning_rate": 2.9201957639747056e-05, + "loss": 0.0254, + "step": 1790 + }, + { + "epoch": 3.0150753768844223, + "grad_norm": 0.12427914887666702, + "learning_rate": 2.9188600288708565e-05, + "loss": 0.0231, + "step": 1800 + }, + { + "epoch": 3.0318257956448913, + "grad_norm": 0.12067142128944397, + "learning_rate": 2.9175135184682053e-05, + "loss": 0.0211, + "step": 1810 + }, + { + "epoch": 3.0485762144053603, + "grad_norm": 0.1151166558265686, + "learning_rate": 2.9161562429926038e-05, + "loss": 0.0228, + "step": 1820 + }, + { + "epoch": 3.0653266331658293, + "grad_norm": 0.12743020057678223, + "learning_rate": 2.9147882127516568e-05, + "loss": 0.0227, + "step": 1830 + }, + { + "epoch": 3.0820770519262983, + "grad_norm": 0.12181981652975082, + "learning_rate": 2.913409438134644e-05, + "loss": 0.024, + "step": 1840 + }, + { + "epoch": 3.0988274706867673, + "grad_norm": 0.1195250153541565, + "learning_rate": 2.9120199296124436e-05, + "loss": 0.0251, + "step": 1850 + }, + { + "epoch": 3.1155778894472363, + "grad_norm": 0.09719762206077576, + "learning_rate": 2.9106196977374477e-05, + "loss": 0.0232, + "step": 1860 + }, + { + "epoch": 3.1323283082077054, + "grad_norm": 0.11931390315294266, + "learning_rate": 2.9092087531434878e-05, + "loss": 0.0236, + "step": 1870 + }, + { + "epoch": 3.1490787269681744, + "grad_norm": 0.10895305871963501, + "learning_rate": 2.9077871065457495e-05, + "loss": 0.023, + "step": 1880 + }, + { + "epoch": 3.1658291457286434, + "grad_norm": 0.1338292807340622, + "learning_rate": 2.9063547687406945e-05, + "loss": 0.024, + "step": 1890 + }, + { + "epoch": 3.1825795644891124, + "grad_norm": 0.11511155962944031, + "learning_rate": 2.9049117506059756e-05, + "loss": 0.0208, + "step": 1900 + }, + { + "epoch": 3.1993299832495814, + "grad_norm": 0.11553452163934708, + "learning_rate": 2.9034580631003564e-05, + "loss": 0.0232, + "step": 1910 + }, + { + "epoch": 3.2160804020100504, + "grad_norm": 0.1008048877120018, + "learning_rate": 2.9019937172636276e-05, + "loss": 0.0222, + "step": 1920 + }, + { + "epoch": 3.2328308207705194, + "grad_norm": 0.10765612125396729, + "learning_rate": 2.9005187242165228e-05, + "loss": 0.0226, + "step": 1930 + }, + { + "epoch": 3.2495812395309884, + "grad_norm": 0.11185599863529205, + "learning_rate": 2.899033095160633e-05, + "loss": 0.0213, + "step": 1940 + }, + { + "epoch": 3.2663316582914574, + "grad_norm": 0.09481639415025711, + "learning_rate": 2.8975368413783235e-05, + "loss": 0.0217, + "step": 1950 + }, + { + "epoch": 3.2830820770519265, + "grad_norm": 0.11557790637016296, + "learning_rate": 2.8960299742326473e-05, + "loss": 0.0231, + "step": 1960 + }, + { + "epoch": 3.2998324958123955, + "grad_norm": 0.13437284529209137, + "learning_rate": 2.8945125051672595e-05, + "loss": 0.0234, + "step": 1970 + }, + { + "epoch": 3.3165829145728645, + "grad_norm": 0.11012782156467438, + "learning_rate": 2.892984445706328e-05, + "loss": 0.0225, + "step": 1980 + }, + { + "epoch": 3.3333333333333335, + "grad_norm": 0.11317654699087143, + "learning_rate": 2.891445807454449e-05, + "loss": 0.024, + "step": 1990 + }, + { + "epoch": 3.3500837520938025, + "grad_norm": 0.19488593935966492, + "learning_rate": 2.8898966020965576e-05, + "loss": 0.024, + "step": 2000 + }, + { + "epoch": 3.3668341708542715, + "grad_norm": 0.10659220814704895, + "learning_rate": 2.8883368413978386e-05, + "loss": 0.0225, + "step": 2010 + }, + { + "epoch": 3.3835845896147405, + "grad_norm": 0.09751500934362411, + "learning_rate": 2.8867665372036375e-05, + "loss": 0.0258, + "step": 2020 + }, + { + "epoch": 3.4003350083752095, + "grad_norm": 0.09632577747106552, + "learning_rate": 2.8851857014393705e-05, + "loss": 0.0222, + "step": 2030 + }, + { + "epoch": 3.4170854271356785, + "grad_norm": 0.12793496251106262, + "learning_rate": 2.8835943461104354e-05, + "loss": 0.0214, + "step": 2040 + }, + { + "epoch": 3.4338358458961475, + "grad_norm": 0.11942226439714432, + "learning_rate": 2.8819924833021176e-05, + "loss": 0.023, + "step": 2050 + }, + { + "epoch": 3.4505862646566166, + "grad_norm": 0.10179294645786285, + "learning_rate": 2.8803801251795007e-05, + "loss": 0.0212, + "step": 2060 + }, + { + "epoch": 3.4673366834170856, + "grad_norm": 0.1228783056139946, + "learning_rate": 2.8787572839873725e-05, + "loss": 0.0225, + "step": 2070 + }, + { + "epoch": 3.4840871021775546, + "grad_norm": 0.13721901178359985, + "learning_rate": 2.8771239720501333e-05, + "loss": 0.0199, + "step": 2080 + }, + { + "epoch": 3.5008375209380236, + "grad_norm": 0.1087748259305954, + "learning_rate": 2.8754802017717015e-05, + "loss": 0.0228, + "step": 2090 + }, + { + "epoch": 3.5175879396984926, + "grad_norm": 0.14415811002254486, + "learning_rate": 2.87382598563542e-05, + "loss": 0.0241, + "step": 2100 + }, + { + "epoch": 3.5343383584589616, + "grad_norm": 0.10477285832166672, + "learning_rate": 2.8721613362039607e-05, + "loss": 0.02, + "step": 2110 + }, + { + "epoch": 3.5510887772194306, + "grad_norm": 0.10786721855401993, + "learning_rate": 2.8704862661192295e-05, + "loss": 0.0212, + "step": 2120 + }, + { + "epoch": 3.5678391959798996, + "grad_norm": 0.10725842416286469, + "learning_rate": 2.86880078810227e-05, + "loss": 0.022, + "step": 2130 + }, + { + "epoch": 3.5845896147403686, + "grad_norm": 0.11172711849212646, + "learning_rate": 2.8671049149531677e-05, + "loss": 0.021, + "step": 2140 + }, + { + "epoch": 3.6013400335008376, + "grad_norm": 0.10756686329841614, + "learning_rate": 2.8653986595509524e-05, + "loss": 0.0208, + "step": 2150 + }, + { + "epoch": 3.6180904522613067, + "grad_norm": 0.08842871338129044, + "learning_rate": 2.8636820348534987e-05, + "loss": 0.0208, + "step": 2160 + }, + { + "epoch": 3.6348408710217757, + "grad_norm": 0.10979699343442917, + "learning_rate": 2.861955053897431e-05, + "loss": 0.0204, + "step": 2170 + }, + { + "epoch": 3.6515912897822447, + "grad_norm": 0.0892506092786789, + "learning_rate": 2.8602177297980213e-05, + "loss": 0.0209, + "step": 2180 + }, + { + "epoch": 3.6683417085427137, + "grad_norm": 0.10864245146512985, + "learning_rate": 2.8584700757490913e-05, + "loss": 0.0214, + "step": 2190 + }, + { + "epoch": 3.6850921273031827, + "grad_norm": 0.12509699165821075, + "learning_rate": 2.8567121050229122e-05, + "loss": 0.0236, + "step": 2200 + }, + { + "epoch": 3.7018425460636517, + "grad_norm": 0.11275574564933777, + "learning_rate": 2.8549438309701028e-05, + "loss": 0.0237, + "step": 2210 + }, + { + "epoch": 3.7185929648241207, + "grad_norm": 0.12549647688865662, + "learning_rate": 2.8531652670195292e-05, + "loss": 0.021, + "step": 2220 + }, + { + "epoch": 3.7353433835845897, + "grad_norm": 0.11705254763364792, + "learning_rate": 2.8513764266782025e-05, + "loss": 0.0224, + "step": 2230 + }, + { + "epoch": 3.7520938023450587, + "grad_norm": 0.16595874726772308, + "learning_rate": 2.849577323531176e-05, + "loss": 0.0224, + "step": 2240 + }, + { + "epoch": 3.7688442211055277, + "grad_norm": 0.10253427922725677, + "learning_rate": 2.8477679712414424e-05, + "loss": 0.0215, + "step": 2250 + }, + { + "epoch": 3.7855946398659968, + "grad_norm": 0.10289912670850754, + "learning_rate": 2.8459483835498298e-05, + "loss": 0.0237, + "step": 2260 + }, + { + "epoch": 3.8023450586264658, + "grad_norm": 0.11400792002677917, + "learning_rate": 2.844118574274896e-05, + "loss": 0.0205, + "step": 2270 + }, + { + "epoch": 3.819095477386935, + "grad_norm": 0.11081373691558838, + "learning_rate": 2.842278557312828e-05, + "loss": 0.0184, + "step": 2280 + }, + { + "epoch": 3.835845896147404, + "grad_norm": 0.13571597635746002, + "learning_rate": 2.84042834663733e-05, + "loss": 0.0207, + "step": 2290 + }, + { + "epoch": 3.852596314907873, + "grad_norm": 0.10395380854606628, + "learning_rate": 2.8385679562995235e-05, + "loss": 0.0209, + "step": 2300 + }, + { + "epoch": 3.869346733668342, + "grad_norm": 0.10459823906421661, + "learning_rate": 2.8366974004278357e-05, + "loss": 0.0242, + "step": 2310 + }, + { + "epoch": 3.886097152428811, + "grad_norm": 0.10516241937875748, + "learning_rate": 2.834816693227895e-05, + "loss": 0.0205, + "step": 2320 + }, + { + "epoch": 3.90284757118928, + "grad_norm": 0.1241426020860672, + "learning_rate": 2.8329258489824234e-05, + "loss": 0.0227, + "step": 2330 + }, + { + "epoch": 3.919597989949749, + "grad_norm": 0.10768191516399384, + "learning_rate": 2.8310248820511248e-05, + "loss": 0.0215, + "step": 2340 + }, + { + "epoch": 3.936348408710218, + "grad_norm": 0.10944578796625137, + "learning_rate": 2.829113806870581e-05, + "loss": 0.0214, + "step": 2350 + }, + { + "epoch": 3.953098827470687, + "grad_norm": 0.11351463198661804, + "learning_rate": 2.8271926379541363e-05, + "loss": 0.022, + "step": 2360 + }, + { + "epoch": 3.969849246231156, + "grad_norm": 0.12451263517141342, + "learning_rate": 2.8252613898917923e-05, + "loss": 0.0216, + "step": 2370 + }, + { + "epoch": 3.986599664991625, + "grad_norm": 0.11138986051082611, + "learning_rate": 2.8233200773500942e-05, + "loss": 0.0215, + "step": 2380 + }, + { + "epoch": 4.0033500837520934, + "grad_norm": 0.14621508121490479, + "learning_rate": 2.8213687150720204e-05, + "loss": 0.0227, + "step": 2390 + }, + { + "epoch": 4.0201005025125625, + "grad_norm": 0.12678438425064087, + "learning_rate": 2.8194073178768703e-05, + "loss": 0.0211, + "step": 2400 + }, + { + "epoch": 4.0368509212730315, + "grad_norm": 0.10121660679578781, + "learning_rate": 2.817435900660152e-05, + "loss": 0.0191, + "step": 2410 + }, + { + "epoch": 4.0536013400335005, + "grad_norm": 0.09031516313552856, + "learning_rate": 2.8154544783934688e-05, + "loss": 0.0241, + "step": 2420 + }, + { + "epoch": 4.0703517587939695, + "grad_norm": 0.12526579201221466, + "learning_rate": 2.8134630661244055e-05, + "loss": 0.0212, + "step": 2430 + }, + { + "epoch": 4.0871021775544385, + "grad_norm": 0.09737418591976166, + "learning_rate": 2.8114616789764146e-05, + "loss": 0.0196, + "step": 2440 + }, + { + "epoch": 4.1038525963149075, + "grad_norm": 0.12097261101007462, + "learning_rate": 2.8094503321487017e-05, + "loss": 0.0208, + "step": 2450 + }, + { + "epoch": 4.1206030150753765, + "grad_norm": 0.10906875133514404, + "learning_rate": 2.8074290409161084e-05, + "loss": 0.0183, + "step": 2460 + }, + { + "epoch": 4.1373534338358455, + "grad_norm": 0.12879160046577454, + "learning_rate": 2.8053978206289984e-05, + "loss": 0.0194, + "step": 2470 + }, + { + "epoch": 4.1541038525963145, + "grad_norm": 0.0820513367652893, + "learning_rate": 2.803356686713139e-05, + "loss": 0.0214, + "step": 2480 + }, + { + "epoch": 4.1708542713567835, + "grad_norm": 0.07784825563430786, + "learning_rate": 2.801305654669587e-05, + "loss": 0.0217, + "step": 2490 + }, + { + "epoch": 4.187604690117253, + "grad_norm": 0.0847000777721405, + "learning_rate": 2.799244740074566e-05, + "loss": 0.0195, + "step": 2500 + }, + { + "epoch": 4.204355108877722, + "grad_norm": 0.10731722414493561, + "learning_rate": 2.797173958579353e-05, + "loss": 0.0206, + "step": 2510 + }, + { + "epoch": 4.221105527638191, + "grad_norm": 0.1059015691280365, + "learning_rate": 2.7950933259101575e-05, + "loss": 0.0189, + "step": 2520 + }, + { + "epoch": 4.23785594639866, + "grad_norm": 0.10778709501028061, + "learning_rate": 2.793002857868001e-05, + "loss": 0.0208, + "step": 2530 + }, + { + "epoch": 4.254606365159129, + "grad_norm": 0.08753228187561035, + "learning_rate": 2.7909025703285983e-05, + "loss": 0.0195, + "step": 2540 + }, + { + "epoch": 4.271356783919598, + "grad_norm": 0.1175403892993927, + "learning_rate": 2.7887924792422385e-05, + "loss": 0.023, + "step": 2550 + }, + { + "epoch": 4.288107202680067, + "grad_norm": 0.12728635966777802, + "learning_rate": 2.78667260063366e-05, + "loss": 0.0215, + "step": 2560 + }, + { + "epoch": 4.304857621440536, + "grad_norm": 0.10965393483638763, + "learning_rate": 2.784542950601931e-05, + "loss": 0.019, + "step": 2570 + }, + { + "epoch": 4.321608040201005, + "grad_norm": 0.10030016303062439, + "learning_rate": 2.7824035453203293e-05, + "loss": 0.0214, + "step": 2580 + }, + { + "epoch": 4.338358458961474, + "grad_norm": 0.10159602761268616, + "learning_rate": 2.7802544010362154e-05, + "loss": 0.018, + "step": 2590 + }, + { + "epoch": 4.355108877721943, + "grad_norm": 0.08524280786514282, + "learning_rate": 2.778095534070912e-05, + "loss": 0.0225, + "step": 2600 + }, + { + "epoch": 4.371859296482412, + "grad_norm": 0.08864463865756989, + "learning_rate": 2.7759269608195777e-05, + "loss": 0.0214, + "step": 2610 + }, + { + "epoch": 4.388609715242881, + "grad_norm": 0.0836586281657219, + "learning_rate": 2.773748697751086e-05, + "loss": 0.019, + "step": 2620 + }, + { + "epoch": 4.40536013400335, + "grad_norm": 0.10019984096288681, + "learning_rate": 2.7715607614078956e-05, + "loss": 0.0204, + "step": 2630 + }, + { + "epoch": 4.422110552763819, + "grad_norm": 0.097859688103199, + "learning_rate": 2.7693631684059305e-05, + "loss": 0.0217, + "step": 2640 + }, + { + "epoch": 4.438860971524288, + "grad_norm": 0.09158801287412643, + "learning_rate": 2.767155935434448e-05, + "loss": 0.0197, + "step": 2650 + }, + { + "epoch": 4.455611390284757, + "grad_norm": 0.09695200622081757, + "learning_rate": 2.7649390792559146e-05, + "loss": 0.0206, + "step": 2660 + }, + { + "epoch": 4.472361809045226, + "grad_norm": 0.09822610765695572, + "learning_rate": 2.7627126167058807e-05, + "loss": 0.0203, + "step": 2670 + }, + { + "epoch": 4.489112227805695, + "grad_norm": 0.09746448695659637, + "learning_rate": 2.7604765646928482e-05, + "loss": 0.0204, + "step": 2680 + }, + { + "epoch": 4.505862646566165, + "grad_norm": 0.10244615375995636, + "learning_rate": 2.758230940198147e-05, + "loss": 0.0215, + "step": 2690 + }, + { + "epoch": 4.522613065326633, + "grad_norm": 0.10013485699892044, + "learning_rate": 2.755975760275802e-05, + "loss": 0.0198, + "step": 2700 + }, + { + "epoch": 4.539363484087103, + "grad_norm": 0.09892448037862778, + "learning_rate": 2.7537110420524057e-05, + "loss": 0.0199, + "step": 2710 + }, + { + "epoch": 4.556113902847571, + "grad_norm": 0.09825485199689865, + "learning_rate": 2.7514368027269875e-05, + "loss": 0.0186, + "step": 2720 + }, + { + "epoch": 4.572864321608041, + "grad_norm": 0.11656782776117325, + "learning_rate": 2.749153059570883e-05, + "loss": 0.0217, + "step": 2730 + }, + { + "epoch": 4.589614740368509, + "grad_norm": 0.08932335674762726, + "learning_rate": 2.7468598299276044e-05, + "loss": 0.0197, + "step": 2740 + }, + { + "epoch": 4.606365159128979, + "grad_norm": 0.09198352694511414, + "learning_rate": 2.7445571312127058e-05, + "loss": 0.0168, + "step": 2750 + }, + { + "epoch": 4.623115577889447, + "grad_norm": 0.09846565872430801, + "learning_rate": 2.7422449809136538e-05, + "loss": 0.0183, + "step": 2760 + }, + { + "epoch": 4.639865996649917, + "grad_norm": 0.07488419115543365, + "learning_rate": 2.739923396589693e-05, + "loss": 0.0209, + "step": 2770 + }, + { + "epoch": 4.656616415410385, + "grad_norm": 0.11510682106018066, + "learning_rate": 2.737592395871713e-05, + "loss": 0.0184, + "step": 2780 + }, + { + "epoch": 4.673366834170855, + "grad_norm": 0.08309126645326614, + "learning_rate": 2.7352519964621146e-05, + "loss": 0.0175, + "step": 2790 + }, + { + "epoch": 4.690117252931323, + "grad_norm": 0.0989244133234024, + "learning_rate": 2.7329022161346758e-05, + "loss": 0.02, + "step": 2800 + }, + { + "epoch": 4.706867671691793, + "grad_norm": 0.09160429984331131, + "learning_rate": 2.730543072734417e-05, + "loss": 0.0211, + "step": 2810 + }, + { + "epoch": 4.723618090452261, + "grad_norm": 0.13737095892429352, + "learning_rate": 2.728174584177463e-05, + "loss": 0.02, + "step": 2820 + }, + { + "epoch": 4.740368509212731, + "grad_norm": 0.09824742376804352, + "learning_rate": 2.725796768450911e-05, + "loss": 0.0187, + "step": 2830 + }, + { + "epoch": 4.757118927973199, + "grad_norm": 0.07674989849328995, + "learning_rate": 2.7234096436126903e-05, + "loss": 0.0201, + "step": 2840 + }, + { + "epoch": 4.773869346733669, + "grad_norm": 0.09987271577119827, + "learning_rate": 2.721013227791428e-05, + "loss": 0.0201, + "step": 2850 + }, + { + "epoch": 4.790619765494137, + "grad_norm": 0.10610336065292358, + "learning_rate": 2.7186075391863088e-05, + "loss": 0.0216, + "step": 2860 + }, + { + "epoch": 4.807370184254607, + "grad_norm": 0.09385973960161209, + "learning_rate": 2.7161925960669383e-05, + "loss": 0.0218, + "step": 2870 + }, + { + "epoch": 4.824120603015075, + "grad_norm": 0.1261572688817978, + "learning_rate": 2.713768416773205e-05, + "loss": 0.0201, + "step": 2880 + }, + { + "epoch": 4.840871021775545, + "grad_norm": 0.06238450109958649, + "learning_rate": 2.7113350197151394e-05, + "loss": 0.0192, + "step": 2890 + }, + { + "epoch": 4.857621440536013, + "grad_norm": 0.08462442457675934, + "learning_rate": 2.708892423372774e-05, + "loss": 0.0189, + "step": 2900 + }, + { + "epoch": 4.874371859296483, + "grad_norm": 0.11602508276700974, + "learning_rate": 2.7064406462960044e-05, + "loss": 0.0175, + "step": 2910 + }, + { + "epoch": 4.891122278056951, + "grad_norm": 0.08591975271701813, + "learning_rate": 2.7039797071044482e-05, + "loss": 0.023, + "step": 2920 + }, + { + "epoch": 4.907872696817421, + "grad_norm": 0.11215902119874954, + "learning_rate": 2.7015096244873028e-05, + "loss": 0.0204, + "step": 2930 + }, + { + "epoch": 4.924623115577889, + "grad_norm": 0.11343713849782944, + "learning_rate": 2.6990304172032038e-05, + "loss": 0.0199, + "step": 2940 + }, + { + "epoch": 4.941373534338359, + "grad_norm": 0.08644749969244003, + "learning_rate": 2.6965421040800823e-05, + "loss": 0.0219, + "step": 2950 + }, + { + "epoch": 4.958123953098827, + "grad_norm": 0.0927010178565979, + "learning_rate": 2.6940447040150228e-05, + "loss": 0.0194, + "step": 2960 + }, + { + "epoch": 4.974874371859297, + "grad_norm": 0.10158302634954453, + "learning_rate": 2.691538235974118e-05, + "loss": 0.0205, + "step": 2970 + }, + { + "epoch": 4.991624790619765, + "grad_norm": 0.10737495869398117, + "learning_rate": 2.6890227189923285e-05, + "loss": 0.0189, + "step": 2980 + }, + { + "epoch": 5.008375209380235, + "grad_norm": 0.07879546284675598, + "learning_rate": 2.6864981721733317e-05, + "loss": 0.0179, + "step": 2990 + }, + { + "epoch": 5.025125628140704, + "grad_norm": 0.10061459243297577, + "learning_rate": 2.683964614689383e-05, + "loss": 0.0196, + "step": 3000 + }, + { + "epoch": 5.041876046901173, + "grad_norm": 0.05866040289402008, + "learning_rate": 2.681422065781169e-05, + "loss": 0.0156, + "step": 3010 + }, + { + "epoch": 5.058626465661642, + "grad_norm": 0.13595989346504211, + "learning_rate": 2.6788705447576563e-05, + "loss": 0.0185, + "step": 3020 + }, + { + "epoch": 5.075376884422111, + "grad_norm": 0.10418418049812317, + "learning_rate": 2.6763100709959524e-05, + "loss": 0.0172, + "step": 3030 + }, + { + "epoch": 5.09212730318258, + "grad_norm": 0.09122584760189056, + "learning_rate": 2.6737406639411537e-05, + "loss": 0.0193, + "step": 3040 + }, + { + "epoch": 5.108877721943049, + "grad_norm": 0.10324976593255997, + "learning_rate": 2.6711623431061985e-05, + "loss": 0.0171, + "step": 3050 + }, + { + "epoch": 5.125628140703517, + "grad_norm": 0.11300911009311676, + "learning_rate": 2.6685751280717195e-05, + "loss": 0.0176, + "step": 3060 + }, + { + "epoch": 5.142378559463987, + "grad_norm": 0.08261276036500931, + "learning_rate": 2.665979038485896e-05, + "loss": 0.0195, + "step": 3070 + }, + { + "epoch": 5.159128978224456, + "grad_norm": 0.10486511141061783, + "learning_rate": 2.6633740940643022e-05, + "loss": 0.017, + "step": 3080 + }, + { + "epoch": 5.175879396984925, + "grad_norm": 0.10391178727149963, + "learning_rate": 2.6607603145897594e-05, + "loss": 0.0201, + "step": 3090 + }, + { + "epoch": 5.192629815745394, + "grad_norm": 0.0923885777592659, + "learning_rate": 2.6581377199121853e-05, + "loss": 0.0191, + "step": 3100 + }, + { + "epoch": 5.209380234505863, + "grad_norm": 0.12680231034755707, + "learning_rate": 2.6555063299484434e-05, + "loss": 0.0177, + "step": 3110 + }, + { + "epoch": 5.226130653266332, + "grad_norm": 0.07475026696920395, + "learning_rate": 2.6528661646821917e-05, + "loss": 0.0193, + "step": 3120 + }, + { + "epoch": 5.242881072026801, + "grad_norm": 0.10191482305526733, + "learning_rate": 2.6502172441637302e-05, + "loss": 0.0193, + "step": 3130 + }, + { + "epoch": 5.259631490787269, + "grad_norm": 0.08599104732275009, + "learning_rate": 2.6475595885098493e-05, + "loss": 0.0193, + "step": 3140 + }, + { + "epoch": 5.276381909547739, + "grad_norm": 0.0979737862944603, + "learning_rate": 2.644893217903677e-05, + "loss": 0.0177, + "step": 3150 + }, + { + "epoch": 5.293132328308207, + "grad_norm": 0.1444755345582962, + "learning_rate": 2.6422181525945265e-05, + "loss": 0.0185, + "step": 3160 + }, + { + "epoch": 5.309882747068677, + "grad_norm": 0.0918390303850174, + "learning_rate": 2.63953441289774e-05, + "loss": 0.02, + "step": 3170 + }, + { + "epoch": 5.326633165829146, + "grad_norm": 0.11050040274858475, + "learning_rate": 2.6368420191945365e-05, + "loss": 0.0181, + "step": 3180 + }, + { + "epoch": 5.343383584589615, + "grad_norm": 0.0900280773639679, + "learning_rate": 2.6341409919318572e-05, + "loss": 0.017, + "step": 3190 + }, + { + "epoch": 5.360134003350084, + "grad_norm": 0.08904273808002472, + "learning_rate": 2.6314313516222084e-05, + "loss": 0.0195, + "step": 3200 + }, + { + "epoch": 5.376884422110553, + "grad_norm": 0.10394366830587387, + "learning_rate": 2.6287131188435075e-05, + "loss": 0.0193, + "step": 3210 + }, + { + "epoch": 5.393634840871022, + "grad_norm": 0.10032975673675537, + "learning_rate": 2.6259863142389252e-05, + "loss": 0.0164, + "step": 3220 + }, + { + "epoch": 5.410385259631491, + "grad_norm": 0.09366985410451889, + "learning_rate": 2.62325095851673e-05, + "loss": 0.0171, + "step": 3230 + }, + { + "epoch": 5.42713567839196, + "grad_norm": 0.08237487822771072, + "learning_rate": 2.620507072450131e-05, + "loss": 0.0191, + "step": 3240 + }, + { + "epoch": 5.443886097152429, + "grad_norm": 0.10653271526098251, + "learning_rate": 2.6177546768771183e-05, + "loss": 0.0173, + "step": 3250 + }, + { + "epoch": 5.460636515912898, + "grad_norm": 0.08800845593214035, + "learning_rate": 2.6149937927003062e-05, + "loss": 0.018, + "step": 3260 + }, + { + "epoch": 5.477386934673367, + "grad_norm": 0.10618974268436432, + "learning_rate": 2.6122244408867766e-05, + "loss": 0.0198, + "step": 3270 + }, + { + "epoch": 5.494137353433836, + "grad_norm": 0.0838366374373436, + "learning_rate": 2.6094466424679148e-05, + "loss": 0.0196, + "step": 3280 + }, + { + "epoch": 5.510887772194305, + "grad_norm": 0.08228679746389389, + "learning_rate": 2.606660418539253e-05, + "loss": 0.0208, + "step": 3290 + }, + { + "epoch": 5.527638190954773, + "grad_norm": 0.10232897847890854, + "learning_rate": 2.6038657902603107e-05, + "loss": 0.0173, + "step": 3300 + }, + { + "epoch": 5.544388609715243, + "grad_norm": 0.0675511434674263, + "learning_rate": 2.6010627788544314e-05, + "loss": 0.0171, + "step": 3310 + }, + { + "epoch": 5.561139028475711, + "grad_norm": 0.11169396340847015, + "learning_rate": 2.5982514056086238e-05, + "loss": 0.0202, + "step": 3320 + }, + { + "epoch": 5.577889447236181, + "grad_norm": 0.09199520200490952, + "learning_rate": 2.5954316918733992e-05, + "loss": 0.0188, + "step": 3330 + }, + { + "epoch": 5.594639865996649, + "grad_norm": 0.09636520594358444, + "learning_rate": 2.5926036590626093e-05, + "loss": 0.0189, + "step": 3340 + }, + { + "epoch": 5.611390284757119, + "grad_norm": 0.1068258136510849, + "learning_rate": 2.589767328653283e-05, + "loss": 0.0185, + "step": 3350 + }, + { + "epoch": 5.628140703517588, + "grad_norm": 0.0738004669547081, + "learning_rate": 2.5869227221854646e-05, + "loss": 0.0153, + "step": 3360 + }, + { + "epoch": 5.644891122278057, + "grad_norm": 0.08421333134174347, + "learning_rate": 2.584069861262049e-05, + "loss": 0.0199, + "step": 3370 + }, + { + "epoch": 5.661641541038526, + "grad_norm": 0.1111551895737648, + "learning_rate": 2.58120876754862e-05, + "loss": 0.0147, + "step": 3380 + }, + { + "epoch": 5.678391959798995, + "grad_norm": 0.0835203304886818, + "learning_rate": 2.5783394627732815e-05, + "loss": 0.0168, + "step": 3390 + }, + { + "epoch": 5.695142378559464, + "grad_norm": 0.12761108577251434, + "learning_rate": 2.5754619687264957e-05, + "loss": 0.0189, + "step": 3400 + }, + { + "epoch": 5.711892797319933, + "grad_norm": 0.0886048823595047, + "learning_rate": 2.5725763072609178e-05, + "loss": 0.0164, + "step": 3410 + }, + { + "epoch": 5.728643216080402, + "grad_norm": 0.09086675196886063, + "learning_rate": 2.569682500291229e-05, + "loss": 0.0151, + "step": 3420 + }, + { + "epoch": 5.745393634840871, + "grad_norm": 0.10654979199171066, + "learning_rate": 2.5667805697939687e-05, + "loss": 0.0173, + "step": 3430 + }, + { + "epoch": 5.76214405360134, + "grad_norm": 0.09250108897686005, + "learning_rate": 2.5638705378073704e-05, + "loss": 0.0207, + "step": 3440 + }, + { + "epoch": 5.778894472361809, + "grad_norm": 0.09105037152767181, + "learning_rate": 2.5609524264311934e-05, + "loss": 0.0175, + "step": 3450 + }, + { + "epoch": 5.795644891122278, + "grad_norm": 0.10493557155132294, + "learning_rate": 2.5580262578265536e-05, + "loss": 0.0217, + "step": 3460 + }, + { + "epoch": 5.812395309882747, + "grad_norm": 0.09852892905473709, + "learning_rate": 2.5550920542157557e-05, + "loss": 0.0153, + "step": 3470 + }, + { + "epoch": 5.8291457286432165, + "grad_norm": 0.09802744537591934, + "learning_rate": 2.5521498378821273e-05, + "loss": 0.0195, + "step": 3480 + }, + { + "epoch": 5.8458961474036855, + "grad_norm": 0.07921155542135239, + "learning_rate": 2.549199631169844e-05, + "loss": 0.0171, + "step": 3490 + }, + { + "epoch": 5.8626465661641545, + "grad_norm": 0.09621422737836838, + "learning_rate": 2.5462414564837663e-05, + "loss": 0.0168, + "step": 3500 + }, + { + "epoch": 5.8793969849246235, + "grad_norm": 0.0989849641919136, + "learning_rate": 2.543275336289264e-05, + "loss": 0.017, + "step": 3510 + }, + { + "epoch": 5.8961474036850925, + "grad_norm": 0.11342565715312958, + "learning_rate": 2.540301293112048e-05, + "loss": 0.018, + "step": 3520 + }, + { + "epoch": 5.9128978224455615, + "grad_norm": 0.1399437040090561, + "learning_rate": 2.537319349538e-05, + "loss": 0.0182, + "step": 3530 + }, + { + "epoch": 5.9296482412060305, + "grad_norm": 0.09442800283432007, + "learning_rate": 2.534329528212999e-05, + "loss": 0.0164, + "step": 3540 + }, + { + "epoch": 5.9463986599664995, + "grad_norm": 0.09357256442308426, + "learning_rate": 2.531331851842751e-05, + "loss": 0.017, + "step": 3550 + }, + { + "epoch": 5.9631490787269685, + "grad_norm": 0.10102587193250656, + "learning_rate": 2.528326343192615e-05, + "loss": 0.0161, + "step": 3560 + }, + { + "epoch": 5.9798994974874375, + "grad_norm": 0.09273260831832886, + "learning_rate": 2.5253130250874316e-05, + "loss": 0.0168, + "step": 3570 + }, + { + "epoch": 5.9966499162479066, + "grad_norm": 0.09276672452688217, + "learning_rate": 2.5222919204113494e-05, + "loss": 0.0198, + "step": 3580 + }, + { + "epoch": 6.013400335008376, + "grad_norm": 0.15660251677036285, + "learning_rate": 2.5192630521076494e-05, + "loss": 0.0171, + "step": 3590 + }, + { + "epoch": 6.030150753768845, + "grad_norm": 0.10578886419534683, + "learning_rate": 2.5162264431785734e-05, + "loss": 0.0199, + "step": 3600 + }, + { + "epoch": 6.046901172529314, + "grad_norm": 0.10759575664997101, + "learning_rate": 2.5131821166851484e-05, + "loss": 0.0167, + "step": 3610 + }, + { + "epoch": 6.063651591289783, + "grad_norm": 0.08797171711921692, + "learning_rate": 2.5101300957470095e-05, + "loss": 0.0177, + "step": 3620 + }, + { + "epoch": 6.080402010050252, + "grad_norm": 0.06793040037155151, + "learning_rate": 2.507070403542228e-05, + "loss": 0.0166, + "step": 3630 + }, + { + "epoch": 6.097152428810721, + "grad_norm": 0.08354511857032776, + "learning_rate": 2.504003063307132e-05, + "loss": 0.0177, + "step": 3640 + }, + { + "epoch": 6.11390284757119, + "grad_norm": 0.06704747676849365, + "learning_rate": 2.500928098336131e-05, + "loss": 0.0184, + "step": 3650 + }, + { + "epoch": 6.130653266331659, + "grad_norm": 0.122153140604496, + "learning_rate": 2.497845531981541e-05, + "loss": 0.0191, + "step": 3660 + }, + { + "epoch": 6.147403685092128, + "grad_norm": 0.08629975467920303, + "learning_rate": 2.4947553876534047e-05, + "loss": 0.0175, + "step": 3670 + }, + { + "epoch": 6.164154103852597, + "grad_norm": 0.09296640753746033, + "learning_rate": 2.4916576888193137e-05, + "loss": 0.0148, + "step": 3680 + }, + { + "epoch": 6.180904522613066, + "grad_norm": 0.06364050507545471, + "learning_rate": 2.488552459004231e-05, + "loss": 0.0169, + "step": 3690 + }, + { + "epoch": 6.197654941373535, + "grad_norm": 0.07617541402578354, + "learning_rate": 2.4854397217903145e-05, + "loss": 0.0176, + "step": 3700 + }, + { + "epoch": 6.214405360134004, + "grad_norm": 0.08891832083463669, + "learning_rate": 2.4823195008167325e-05, + "loss": 0.0172, + "step": 3710 + }, + { + "epoch": 6.231155778894473, + "grad_norm": 0.06869089603424072, + "learning_rate": 2.4791918197794907e-05, + "loss": 0.0161, + "step": 3720 + }, + { + "epoch": 6.247906197654942, + "grad_norm": 0.10788177698850632, + "learning_rate": 2.4760567024312476e-05, + "loss": 0.0154, + "step": 3730 + }, + { + "epoch": 6.264656616415411, + "grad_norm": 0.14439617097377777, + "learning_rate": 2.4729141725811344e-05, + "loss": 0.0183, + "step": 3740 + }, + { + "epoch": 6.28140703517588, + "grad_norm": 0.10388529300689697, + "learning_rate": 2.4697642540945774e-05, + "loss": 0.0194, + "step": 3750 + }, + { + "epoch": 6.298157453936349, + "grad_norm": 0.08002973347902298, + "learning_rate": 2.4666069708931136e-05, + "loss": 0.015, + "step": 3760 + }, + { + "epoch": 6.314907872696818, + "grad_norm": 0.0883835032582283, + "learning_rate": 2.46344234695421e-05, + "loss": 0.0183, + "step": 3770 + }, + { + "epoch": 6.331658291457287, + "grad_norm": 0.08702576160430908, + "learning_rate": 2.460270406311082e-05, + "loss": 0.0159, + "step": 3780 + }, + { + "epoch": 6.348408710217756, + "grad_norm": 0.09502051025629044, + "learning_rate": 2.457091173052511e-05, + "loss": 0.0173, + "step": 3790 + }, + { + "epoch": 6.365159128978225, + "grad_norm": 0.09019004553556442, + "learning_rate": 2.4539046713226593e-05, + "loss": 0.0192, + "step": 3800 + }, + { + "epoch": 6.381909547738694, + "grad_norm": 0.10312219709157944, + "learning_rate": 2.4507109253208898e-05, + "loss": 0.0154, + "step": 3810 + }, + { + "epoch": 6.398659966499163, + "grad_norm": 0.08851608633995056, + "learning_rate": 2.4475099593015813e-05, + "loss": 0.0172, + "step": 3820 + }, + { + "epoch": 6.415410385259632, + "grad_norm": 0.0891028419137001, + "learning_rate": 2.4443017975739423e-05, + "loss": 0.0188, + "step": 3830 + }, + { + "epoch": 6.432160804020101, + "grad_norm": 0.08572867512702942, + "learning_rate": 2.4410864645018288e-05, + "loss": 0.0165, + "step": 3840 + }, + { + "epoch": 6.44891122278057, + "grad_norm": 0.09533847123384476, + "learning_rate": 2.4378639845035583e-05, + "loss": 0.0158, + "step": 3850 + }, + { + "epoch": 6.465661641541039, + "grad_norm": 0.0869334265589714, + "learning_rate": 2.4346343820517237e-05, + "loss": 0.0158, + "step": 3860 + }, + { + "epoch": 6.482412060301508, + "grad_norm": 0.10855986922979355, + "learning_rate": 2.4313976816730095e-05, + "loss": 0.0179, + "step": 3870 + }, + { + "epoch": 6.499162479061977, + "grad_norm": 0.0890968069434166, + "learning_rate": 2.4281539079480047e-05, + "loss": 0.0166, + "step": 3880 + }, + { + "epoch": 6.515912897822446, + "grad_norm": 0.09087546169757843, + "learning_rate": 2.4249030855110128e-05, + "loss": 0.0156, + "step": 3890 + }, + { + "epoch": 6.532663316582915, + "grad_norm": 0.07741452008485794, + "learning_rate": 2.42164523904987e-05, + "loss": 0.0163, + "step": 3900 + }, + { + "epoch": 6.549413735343384, + "grad_norm": 0.09288477897644043, + "learning_rate": 2.4183803933057554e-05, + "loss": 0.0165, + "step": 3910 + }, + { + "epoch": 6.566164154103853, + "grad_norm": 0.08580487966537476, + "learning_rate": 2.4151085730730013e-05, + "loss": 0.0201, + "step": 3920 + }, + { + "epoch": 6.582914572864322, + "grad_norm": 0.079682357609272, + "learning_rate": 2.4118298031989084e-05, + "loss": 0.0163, + "step": 3930 + }, + { + "epoch": 6.599664991624791, + "grad_norm": 0.07965271174907684, + "learning_rate": 2.4085441085835534e-05, + "loss": 0.0162, + "step": 3940 + }, + { + "epoch": 6.61641541038526, + "grad_norm": 0.10382688790559769, + "learning_rate": 2.405251514179604e-05, + "loss": 0.0172, + "step": 3950 + }, + { + "epoch": 6.633165829145729, + "grad_norm": 0.0887349396944046, + "learning_rate": 2.401952044992126e-05, + "loss": 0.0181, + "step": 3960 + }, + { + "epoch": 6.649916247906198, + "grad_norm": 0.10867097228765488, + "learning_rate": 2.3986457260783942e-05, + "loss": 0.016, + "step": 3970 + }, + { + "epoch": 6.666666666666667, + "grad_norm": 0.10828111320734024, + "learning_rate": 2.395332582547703e-05, + "loss": 0.0216, + "step": 3980 + }, + { + "epoch": 6.683417085427136, + "grad_norm": 0.08616909384727478, + "learning_rate": 2.3920126395611755e-05, + "loss": 0.019, + "step": 3990 + }, + { + "epoch": 6.700167504187605, + "grad_norm": 0.10397583991289139, + "learning_rate": 2.3886859223315725e-05, + "loss": 0.0199, + "step": 4000 + }, + { + "epoch": 6.716917922948074, + "grad_norm": 0.12109193205833435, + "learning_rate": 2.385352456123099e-05, + "loss": 0.0191, + "step": 4010 + }, + { + "epoch": 6.733668341708543, + "grad_norm": 0.07098208367824554, + "learning_rate": 2.3820122662512155e-05, + "loss": 0.0164, + "step": 4020 + }, + { + "epoch": 6.750418760469012, + "grad_norm": 0.1095917746424675, + "learning_rate": 2.3786653780824443e-05, + "loss": 0.0153, + "step": 4030 + }, + { + "epoch": 6.767169179229481, + "grad_norm": 0.08910483866930008, + "learning_rate": 2.3753118170341762e-05, + "loss": 0.0168, + "step": 4040 + }, + { + "epoch": 6.78391959798995, + "grad_norm": 0.07048118114471436, + "learning_rate": 2.3719516085744782e-05, + "loss": 0.0152, + "step": 4050 + }, + { + "epoch": 6.800670016750419, + "grad_norm": 0.1071673035621643, + "learning_rate": 2.3685847782218997e-05, + "loss": 0.0158, + "step": 4060 + }, + { + "epoch": 6.817420435510888, + "grad_norm": 0.10280796140432358, + "learning_rate": 2.3652113515452792e-05, + "loss": 0.0173, + "step": 4070 + }, + { + "epoch": 6.834170854271357, + "grad_norm": 0.10277356952428818, + "learning_rate": 2.3618313541635506e-05, + "loss": 0.0173, + "step": 4080 + }, + { + "epoch": 6.850921273031826, + "grad_norm": 0.09373802691698074, + "learning_rate": 2.358444811745547e-05, + "loss": 0.0166, + "step": 4090 + }, + { + "epoch": 6.867671691792295, + "grad_norm": 0.09374375641345978, + "learning_rate": 2.3550517500098068e-05, + "loss": 0.0171, + "step": 4100 + }, + { + "epoch": 6.884422110552764, + "grad_norm": 0.08589532971382141, + "learning_rate": 2.3516521947243787e-05, + "loss": 0.0177, + "step": 4110 + }, + { + "epoch": 6.901172529313233, + "grad_norm": 0.09759582579135895, + "learning_rate": 2.348246171706626e-05, + "loss": 0.0161, + "step": 4120 + }, + { + "epoch": 6.917922948073702, + "grad_norm": 0.08552340418100357, + "learning_rate": 2.3448337068230288e-05, + "loss": 0.0156, + "step": 4130 + }, + { + "epoch": 6.934673366834171, + "grad_norm": 0.07144278287887573, + "learning_rate": 2.3414148259889896e-05, + "loss": 0.0184, + "step": 4140 + }, + { + "epoch": 6.95142378559464, + "grad_norm": 0.08075377345085144, + "learning_rate": 2.3379895551686358e-05, + "loss": 0.0182, + "step": 4150 + }, + { + "epoch": 6.968174204355109, + "grad_norm": 0.09423919767141342, + "learning_rate": 2.334557920374623e-05, + "loss": 0.0173, + "step": 4160 + }, + { + "epoch": 6.984924623115578, + "grad_norm": 0.09013880789279938, + "learning_rate": 2.3311199476679367e-05, + "loss": 0.0172, + "step": 4170 + }, + { + "epoch": 7.001675041876047, + "grad_norm": 0.1004585400223732, + "learning_rate": 2.327675663157693e-05, + "loss": 0.0171, + "step": 4180 + }, + { + "epoch": 7.018425460636516, + "grad_norm": 0.09630786627531052, + "learning_rate": 2.3242250930009446e-05, + "loss": 0.0169, + "step": 4190 + }, + { + "epoch": 7.035175879396985, + "grad_norm": 0.09441886842250824, + "learning_rate": 2.3207682634024777e-05, + "loss": 0.016, + "step": 4200 + }, + { + "epoch": 7.051926298157454, + "grad_norm": 0.1110304594039917, + "learning_rate": 2.3173052006146155e-05, + "loss": 0.0163, + "step": 4210 + }, + { + "epoch": 7.068676716917923, + "grad_norm": 0.07521197199821472, + "learning_rate": 2.313835930937018e-05, + "loss": 0.0146, + "step": 4220 + }, + { + "epoch": 7.085427135678392, + "grad_norm": 0.06828752905130386, + "learning_rate": 2.310360480716483e-05, + "loss": 0.0156, + "step": 4230 + }, + { + "epoch": 7.102177554438861, + "grad_norm": 0.08863577991724014, + "learning_rate": 2.3068788763467434e-05, + "loss": 0.0154, + "step": 4240 + }, + { + "epoch": 7.11892797319933, + "grad_norm": 0.10143563151359558, + "learning_rate": 2.3033911442682708e-05, + "loss": 0.0171, + "step": 4250 + }, + { + "epoch": 7.135678391959799, + "grad_norm": 0.10039176791906357, + "learning_rate": 2.299897310968072e-05, + "loss": 0.0185, + "step": 4260 + }, + { + "epoch": 7.152428810720268, + "grad_norm": 0.06830760836601257, + "learning_rate": 2.2963974029794893e-05, + "loss": 0.0135, + "step": 4270 + }, + { + "epoch": 7.169179229480737, + "grad_norm": 0.06696446985006332, + "learning_rate": 2.2928914468819968e-05, + "loss": 0.016, + "step": 4280 + }, + { + "epoch": 7.185929648241206, + "grad_norm": 0.06527981907129288, + "learning_rate": 2.2893794693010002e-05, + "loss": 0.0149, + "step": 4290 + }, + { + "epoch": 7.202680067001675, + "grad_norm": 0.08876840025186539, + "learning_rate": 2.2858614969076358e-05, + "loss": 0.0179, + "step": 4300 + }, + { + "epoch": 7.219430485762144, + "grad_norm": 0.08040497452020645, + "learning_rate": 2.2823375564185645e-05, + "loss": 0.0172, + "step": 4310 + }, + { + "epoch": 7.236180904522613, + "grad_norm": 0.06119128316640854, + "learning_rate": 2.278807674595773e-05, + "loss": 0.0157, + "step": 4320 + }, + { + "epoch": 7.252931323283082, + "grad_norm": 0.07552856206893921, + "learning_rate": 2.275271878246367e-05, + "loss": 0.0161, + "step": 4330 + }, + { + "epoch": 7.269681742043551, + "grad_norm": 0.09174931794404984, + "learning_rate": 2.271730194222369e-05, + "loss": 0.0157, + "step": 4340 + }, + { + "epoch": 7.28643216080402, + "grad_norm": 0.10719872266054153, + "learning_rate": 2.2681826494205156e-05, + "loss": 0.0175, + "step": 4350 + }, + { + "epoch": 7.303182579564489, + "grad_norm": 0.08944938331842422, + "learning_rate": 2.2646292707820513e-05, + "loss": 0.0146, + "step": 4360 + }, + { + "epoch": 7.319932998324958, + "grad_norm": 0.08042101562023163, + "learning_rate": 2.2610700852925254e-05, + "loss": 0.0145, + "step": 4370 + }, + { + "epoch": 7.336683417085427, + "grad_norm": 0.10604419559240341, + "learning_rate": 2.2575051199815854e-05, + "loss": 0.0175, + "step": 4380 + }, + { + "epoch": 7.353433835845896, + "grad_norm": 0.07255196571350098, + "learning_rate": 2.253934401922773e-05, + "loss": 0.0161, + "step": 4390 + }, + { + "epoch": 7.370184254606365, + "grad_norm": 0.07271875441074371, + "learning_rate": 2.250357958233319e-05, + "loss": 0.0181, + "step": 4400 + }, + { + "epoch": 7.386934673366834, + "grad_norm": 0.08586633205413818, + "learning_rate": 2.2467758160739352e-05, + "loss": 0.0159, + "step": 4410 + }, + { + "epoch": 7.403685092127303, + "grad_norm": 0.07192263007164001, + "learning_rate": 2.2431880026486114e-05, + "loss": 0.0176, + "step": 4420 + }, + { + "epoch": 7.420435510887772, + "grad_norm": 0.0814385861158371, + "learning_rate": 2.2395945452044045e-05, + "loss": 0.0171, + "step": 4430 + }, + { + "epoch": 7.437185929648241, + "grad_norm": 0.0670303925871849, + "learning_rate": 2.2359954710312363e-05, + "loss": 0.0145, + "step": 4440 + }, + { + "epoch": 7.45393634840871, + "grad_norm": 0.08451966196298599, + "learning_rate": 2.232390807461682e-05, + "loss": 0.0148, + "step": 4450 + }, + { + "epoch": 7.4706867671691795, + "grad_norm": 0.09510994702577591, + "learning_rate": 2.2287805818707658e-05, + "loss": 0.016, + "step": 4460 + }, + { + "epoch": 7.4874371859296485, + "grad_norm": 0.08456354588270187, + "learning_rate": 2.2251648216757514e-05, + "loss": 0.014, + "step": 4470 + }, + { + "epoch": 7.5041876046901175, + "grad_norm": 0.11731875687837601, + "learning_rate": 2.2215435543359337e-05, + "loss": 0.0154, + "step": 4480 + }, + { + "epoch": 7.5209380234505865, + "grad_norm": 0.10068397969007492, + "learning_rate": 2.2179168073524308e-05, + "loss": 0.0161, + "step": 4490 + }, + { + "epoch": 7.5376884422110555, + "grad_norm": 0.09683054685592651, + "learning_rate": 2.2142846082679763e-05, + "loss": 0.0168, + "step": 4500 + }, + { + "epoch": 7.5544388609715245, + "grad_norm": 0.06540673226118088, + "learning_rate": 2.2106469846667078e-05, + "loss": 0.013, + "step": 4510 + }, + { + "epoch": 7.5711892797319935, + "grad_norm": 0.0816856399178505, + "learning_rate": 2.2070039641739578e-05, + "loss": 0.0154, + "step": 4520 + }, + { + "epoch": 7.5879396984924625, + "grad_norm": 0.07817541062831879, + "learning_rate": 2.2033555744560462e-05, + "loss": 0.017, + "step": 4530 + }, + { + "epoch": 7.6046901172529315, + "grad_norm": 0.08995489031076431, + "learning_rate": 2.1997018432200674e-05, + "loss": 0.0153, + "step": 4540 + }, + { + "epoch": 7.6214405360134005, + "grad_norm": 0.09150712937116623, + "learning_rate": 2.1960427982136818e-05, + "loss": 0.0154, + "step": 4550 + }, + { + "epoch": 7.63819095477387, + "grad_norm": 0.11141885071992874, + "learning_rate": 2.1923784672249038e-05, + "loss": 0.0159, + "step": 4560 + }, + { + "epoch": 7.654941373534339, + "grad_norm": 0.09261714667081833, + "learning_rate": 2.1887088780818918e-05, + "loss": 0.015, + "step": 4570 + }, + { + "epoch": 7.671691792294808, + "grad_norm": 0.11019844561815262, + "learning_rate": 2.1850340586527354e-05, + "loss": 0.0181, + "step": 4580 + }, + { + "epoch": 7.688442211055277, + "grad_norm": 0.09286614507436752, + "learning_rate": 2.1813540368452468e-05, + "loss": 0.0165, + "step": 4590 + }, + { + "epoch": 7.705192629815746, + "grad_norm": 0.07964478433132172, + "learning_rate": 2.1776688406067447e-05, + "loss": 0.0162, + "step": 4600 + }, + { + "epoch": 7.721943048576215, + "grad_norm": 0.09548459202051163, + "learning_rate": 2.1739784979238447e-05, + "loss": 0.0165, + "step": 4610 + }, + { + "epoch": 7.738693467336684, + "grad_norm": 0.06440986692905426, + "learning_rate": 2.1702830368222478e-05, + "loss": 0.0141, + "step": 4620 + }, + { + "epoch": 7.755443886097153, + "grad_norm": 0.07117027044296265, + "learning_rate": 2.166582485366523e-05, + "loss": 0.0147, + "step": 4630 + }, + { + "epoch": 7.772194304857622, + "grad_norm": 0.08931209146976471, + "learning_rate": 2.1628768716599e-05, + "loss": 0.0147, + "step": 4640 + }, + { + "epoch": 7.788944723618091, + "grad_norm": 0.09026764333248138, + "learning_rate": 2.1591662238440515e-05, + "loss": 0.0159, + "step": 4650 + }, + { + "epoch": 7.80569514237856, + "grad_norm": 0.1498933583498001, + "learning_rate": 2.1554505700988813e-05, + "loss": 0.0155, + "step": 4660 + }, + { + "epoch": 7.822445561139029, + "grad_norm": 0.06725804507732391, + "learning_rate": 2.1517299386423094e-05, + "loss": 0.016, + "step": 4670 + }, + { + "epoch": 7.839195979899498, + "grad_norm": 0.08815880864858627, + "learning_rate": 2.148004357730059e-05, + "loss": 0.0151, + "step": 4680 + }, + { + "epoch": 7.855946398659967, + "grad_norm": 0.06878040730953217, + "learning_rate": 2.1442738556554404e-05, + "loss": 0.0172, + "step": 4690 + }, + { + "epoch": 7.872696817420436, + "grad_norm": 0.06804270297288895, + "learning_rate": 2.1405384607491375e-05, + "loss": 0.0156, + "step": 4700 + }, + { + "epoch": 7.889447236180905, + "grad_norm": 0.070171058177948, + "learning_rate": 2.1367982013789913e-05, + "loss": 0.0153, + "step": 4710 + }, + { + "epoch": 7.906197654941374, + "grad_norm": 0.07167346775531769, + "learning_rate": 2.133053105949786e-05, + "loss": 0.0156, + "step": 4720 + }, + { + "epoch": 7.922948073701843, + "grad_norm": 0.09849503636360168, + "learning_rate": 2.129303202903032e-05, + "loss": 0.0165, + "step": 4730 + }, + { + "epoch": 7.939698492462312, + "grad_norm": 0.09628047794103622, + "learning_rate": 2.1255485207167504e-05, + "loss": 0.016, + "step": 4740 + }, + { + "epoch": 7.956448911222781, + "grad_norm": 0.07518996298313141, + "learning_rate": 2.1217890879052556e-05, + "loss": 0.0148, + "step": 4750 + }, + { + "epoch": 7.97319932998325, + "grad_norm": 0.09254742413759232, + "learning_rate": 2.1180249330189423e-05, + "loss": 0.0177, + "step": 4760 + }, + { + "epoch": 7.989949748743719, + "grad_norm": 0.10075143724679947, + "learning_rate": 2.1142560846440645e-05, + "loss": 0.0173, + "step": 4770 + }, + { + "epoch": 8.006700167504187, + "grad_norm": 0.0872223973274231, + "learning_rate": 2.1104825714025203e-05, + "loss": 0.0155, + "step": 4780 + }, + { + "epoch": 8.023450586264657, + "grad_norm": 0.07023284584283829, + "learning_rate": 2.106704421951634e-05, + "loss": 0.0169, + "step": 4790 + }, + { + "epoch": 8.040201005025125, + "grad_norm": 0.0736195370554924, + "learning_rate": 2.1029216649839404e-05, + "loss": 0.0153, + "step": 4800 + }, + { + "epoch": 8.056951423785595, + "grad_norm": 0.12984727323055267, + "learning_rate": 2.099134329226963e-05, + "loss": 0.0194, + "step": 4810 + }, + { + "epoch": 8.073701842546063, + "grad_norm": 0.06633501499891281, + "learning_rate": 2.0953424434430006e-05, + "loss": 0.0172, + "step": 4820 + }, + { + "epoch": 8.090452261306533, + "grad_norm": 0.07725346088409424, + "learning_rate": 2.091546036428905e-05, + "loss": 0.0167, + "step": 4830 + }, + { + "epoch": 8.107202680067001, + "grad_norm": 0.0776301696896553, + "learning_rate": 2.0877451370158633e-05, + "loss": 0.0167, + "step": 4840 + }, + { + "epoch": 8.12395309882747, + "grad_norm": 0.08715241402387619, + "learning_rate": 2.083939774069181e-05, + "loss": 0.0152, + "step": 4850 + }, + { + "epoch": 8.140703517587939, + "grad_norm": 0.05368698760867119, + "learning_rate": 2.080129976488059e-05, + "loss": 0.0158, + "step": 4860 + }, + { + "epoch": 8.157453936348409, + "grad_norm": 0.07227354496717453, + "learning_rate": 2.0763157732053795e-05, + "loss": 0.0167, + "step": 4870 + }, + { + "epoch": 8.174204355108877, + "grad_norm": 0.08462859690189362, + "learning_rate": 2.0724971931874796e-05, + "loss": 0.0155, + "step": 4880 + }, + { + "epoch": 8.190954773869347, + "grad_norm": 0.0951157882809639, + "learning_rate": 2.0686742654339363e-05, + "loss": 0.0151, + "step": 4890 + }, + { + "epoch": 8.207705192629815, + "grad_norm": 0.07463134825229645, + "learning_rate": 2.0648470189773455e-05, + "loss": 0.0136, + "step": 4900 + }, + { + "epoch": 8.224455611390285, + "grad_norm": 0.08849921822547913, + "learning_rate": 2.0610154828830997e-05, + "loss": 0.0145, + "step": 4910 + }, + { + "epoch": 8.241206030150753, + "grad_norm": 0.08144073933362961, + "learning_rate": 2.0571796862491686e-05, + "loss": 0.0173, + "step": 4920 + }, + { + "epoch": 8.257956448911223, + "grad_norm": 0.08935582637786865, + "learning_rate": 2.0533396582058783e-05, + "loss": 0.0173, + "step": 4930 + }, + { + "epoch": 8.274706867671691, + "grad_norm": 0.06497425585985184, + "learning_rate": 2.0494954279156893e-05, + "loss": 0.0159, + "step": 4940 + }, + { + "epoch": 8.291457286432161, + "grad_norm": 0.09040762484073639, + "learning_rate": 2.0456470245729748e-05, + "loss": 0.018, + "step": 4950 + }, + { + "epoch": 8.308207705192629, + "grad_norm": 0.11972896009683609, + "learning_rate": 2.0417944774038013e-05, + "loss": 0.016, + "step": 4960 + }, + { + "epoch": 8.324958123953099, + "grad_norm": 0.09560854732990265, + "learning_rate": 2.0379378156657026e-05, + "loss": 0.0151, + "step": 4970 + }, + { + "epoch": 8.341708542713567, + "grad_norm": 0.09106440097093582, + "learning_rate": 2.0340770686474618e-05, + "loss": 0.0169, + "step": 4980 + }, + { + "epoch": 8.358458961474037, + "grad_norm": 0.07660805433988571, + "learning_rate": 2.030212265668886e-05, + "loss": 0.0141, + "step": 4990 + }, + { + "epoch": 8.375209380234505, + "grad_norm": 0.08468059450387955, + "learning_rate": 2.0263434360805847e-05, + "loss": 0.0167, + "step": 5000 + }, + { + "epoch": 8.391959798994975, + "grad_norm": 0.0799565464258194, + "learning_rate": 2.0224706092637472e-05, + "loss": 0.0153, + "step": 5010 + }, + { + "epoch": 8.408710217755443, + "grad_norm": 0.10247821360826492, + "learning_rate": 2.018593814629919e-05, + "loss": 0.0168, + "step": 5020 + }, + { + "epoch": 8.425460636515913, + "grad_norm": 0.07618727535009384, + "learning_rate": 2.0147130816207786e-05, + "loss": 0.016, + "step": 5030 + }, + { + "epoch": 8.442211055276381, + "grad_norm": 0.11073528230190277, + "learning_rate": 2.0108284397079127e-05, + "loss": 0.015, + "step": 5040 + }, + { + "epoch": 8.458961474036851, + "grad_norm": 0.10041352361440659, + "learning_rate": 2.006939918392596e-05, + "loss": 0.0156, + "step": 5050 + }, + { + "epoch": 8.47571189279732, + "grad_norm": 0.06778070330619812, + "learning_rate": 2.0030475472055625e-05, + "loss": 0.0151, + "step": 5060 + }, + { + "epoch": 8.492462311557789, + "grad_norm": 0.07777587324380875, + "learning_rate": 1.9991513557067835e-05, + "loss": 0.0144, + "step": 5070 + }, + { + "epoch": 8.509212730318257, + "grad_norm": 0.06961089372634888, + "learning_rate": 1.9952513734852444e-05, + "loss": 0.0139, + "step": 5080 + }, + { + "epoch": 8.525963149078727, + "grad_norm": 0.07862704992294312, + "learning_rate": 1.991347630158717e-05, + "loss": 0.0148, + "step": 5090 + }, + { + "epoch": 8.542713567839195, + "grad_norm": 0.10534326732158661, + "learning_rate": 1.9874401553735385e-05, + "loss": 0.0161, + "step": 5100 + }, + { + "epoch": 8.559463986599665, + "grad_norm": 0.08768954873085022, + "learning_rate": 1.9835289788043814e-05, + "loss": 0.0137, + "step": 5110 + }, + { + "epoch": 8.576214405360133, + "grad_norm": 0.06274593621492386, + "learning_rate": 1.979614130154032e-05, + "loss": 0.0149, + "step": 5120 + }, + { + "epoch": 8.592964824120603, + "grad_norm": 0.0664563998579979, + "learning_rate": 1.9756956391531646e-05, + "loss": 0.013, + "step": 5130 + }, + { + "epoch": 8.609715242881071, + "grad_norm": 0.061676476150751114, + "learning_rate": 1.9717735355601126e-05, + "loss": 0.0162, + "step": 5140 + }, + { + "epoch": 8.626465661641541, + "grad_norm": 0.08172836154699326, + "learning_rate": 1.9678478491606462e-05, + "loss": 0.0174, + "step": 5150 + }, + { + "epoch": 8.64321608040201, + "grad_norm": 0.05784829333424568, + "learning_rate": 1.9639186097677442e-05, + "loss": 0.0149, + "step": 5160 + }, + { + "epoch": 8.65996649916248, + "grad_norm": 0.08542623370885849, + "learning_rate": 1.9599858472213682e-05, + "loss": 0.0147, + "step": 5170 + }, + { + "epoch": 8.676716917922947, + "grad_norm": 0.07957877963781357, + "learning_rate": 1.9560495913882353e-05, + "loss": 0.0133, + "step": 5180 + }, + { + "epoch": 8.693467336683417, + "grad_norm": 0.0810636356472969, + "learning_rate": 1.9521098721615923e-05, + "loss": 0.0145, + "step": 5190 + }, + { + "epoch": 8.710217755443885, + "grad_norm": 0.07732214033603668, + "learning_rate": 1.9481667194609882e-05, + "loss": 0.0145, + "step": 5200 + }, + { + "epoch": 8.726968174204355, + "grad_norm": 0.0780918151140213, + "learning_rate": 1.9442201632320464e-05, + "loss": 0.0147, + "step": 5210 + }, + { + "epoch": 8.743718592964823, + "grad_norm": 0.09253431856632233, + "learning_rate": 1.9402702334462383e-05, + "loss": 0.0155, + "step": 5220 + }, + { + "epoch": 8.760469011725293, + "grad_norm": 0.06418832391500473, + "learning_rate": 1.9363169601006552e-05, + "loss": 0.0126, + "step": 5230 + }, + { + "epoch": 8.777219430485761, + "grad_norm": 0.09778792411088943, + "learning_rate": 1.9323603732177798e-05, + "loss": 0.0146, + "step": 5240 + }, + { + "epoch": 8.793969849246231, + "grad_norm": 0.0891941487789154, + "learning_rate": 1.9284005028452607e-05, + "loss": 0.0141, + "step": 5250 + }, + { + "epoch": 8.8107202680067, + "grad_norm": 0.1019863709807396, + "learning_rate": 1.9244373790556803e-05, + "loss": 0.0154, + "step": 5260 + }, + { + "epoch": 8.82747068676717, + "grad_norm": 0.08988063782453537, + "learning_rate": 1.9204710319463297e-05, + "loss": 0.0151, + "step": 5270 + }, + { + "epoch": 8.844221105527637, + "grad_norm": 0.0668616071343422, + "learning_rate": 1.9165014916389792e-05, + "loss": 0.0128, + "step": 5280 + }, + { + "epoch": 8.860971524288107, + "grad_norm": 0.08318133652210236, + "learning_rate": 1.9125287882796483e-05, + "loss": 0.0159, + "step": 5290 + }, + { + "epoch": 8.877721943048575, + "grad_norm": 0.07382380962371826, + "learning_rate": 1.9085529520383783e-05, + "loss": 0.0142, + "step": 5300 + }, + { + "epoch": 8.894472361809045, + "grad_norm": 0.07555682957172394, + "learning_rate": 1.9045740131090037e-05, + "loss": 0.0142, + "step": 5310 + }, + { + "epoch": 8.911222780569513, + "grad_norm": 0.09456376731395721, + "learning_rate": 1.9005920017089196e-05, + "loss": 0.0139, + "step": 5320 + }, + { + "epoch": 8.927973199329983, + "grad_norm": 0.11305142194032669, + "learning_rate": 1.8966069480788562e-05, + "loss": 0.0146, + "step": 5330 + }, + { + "epoch": 8.944723618090451, + "grad_norm": 0.06259690970182419, + "learning_rate": 1.8926188824826477e-05, + "loss": 0.0154, + "step": 5340 + }, + { + "epoch": 8.961474036850921, + "grad_norm": 0.10811670124530792, + "learning_rate": 1.8886278352070002e-05, + "loss": 0.014, + "step": 5350 + }, + { + "epoch": 8.97822445561139, + "grad_norm": 0.06475009024143219, + "learning_rate": 1.884633836561265e-05, + "loss": 0.014, + "step": 5360 + }, + { + "epoch": 8.99497487437186, + "grad_norm": 0.08297079056501389, + "learning_rate": 1.8806369168772086e-05, + "loss": 0.0141, + "step": 5370 + }, + { + "epoch": 9.011725293132328, + "grad_norm": 0.079750195145607, + "learning_rate": 1.8766371065087778e-05, + "loss": 0.0142, + "step": 5380 + }, + { + "epoch": 9.028475711892797, + "grad_norm": 0.07751394808292389, + "learning_rate": 1.8726344358318736e-05, + "loss": 0.0155, + "step": 5390 + }, + { + "epoch": 9.045226130653266, + "grad_norm": 0.0738295242190361, + "learning_rate": 1.86862893524412e-05, + "loss": 0.0142, + "step": 5400 + }, + { + "epoch": 9.061976549413735, + "grad_norm": 0.11355958878993988, + "learning_rate": 1.864620635164631e-05, + "loss": 0.0161, + "step": 5410 + }, + { + "epoch": 9.078726968174204, + "grad_norm": 0.08234552294015884, + "learning_rate": 1.8606095660337813e-05, + "loss": 0.016, + "step": 5420 + }, + { + "epoch": 9.095477386934673, + "grad_norm": 0.09014493972063065, + "learning_rate": 1.8565957583129753e-05, + "loss": 0.0133, + "step": 5430 + }, + { + "epoch": 9.112227805695142, + "grad_norm": 0.08228293806314468, + "learning_rate": 1.852579242484414e-05, + "loss": 0.0152, + "step": 5440 + }, + { + "epoch": 9.128978224455611, + "grad_norm": 0.06658344715833664, + "learning_rate": 1.848560049050866e-05, + "loss": 0.0158, + "step": 5450 + }, + { + "epoch": 9.14572864321608, + "grad_norm": 0.08792341500520706, + "learning_rate": 1.8445382085354336e-05, + "loss": 0.0122, + "step": 5460 + }, + { + "epoch": 9.16247906197655, + "grad_norm": 0.07732446491718292, + "learning_rate": 1.8405137514813224e-05, + "loss": 0.0141, + "step": 5470 + }, + { + "epoch": 9.179229480737018, + "grad_norm": 0.06328639388084412, + "learning_rate": 1.8364867084516084e-05, + "loss": 0.0138, + "step": 5480 + }, + { + "epoch": 9.195979899497488, + "grad_norm": 0.05957376956939697, + "learning_rate": 1.8324571100290064e-05, + "loss": 0.0137, + "step": 5490 + }, + { + "epoch": 9.212730318257956, + "grad_norm": 0.0765262320637703, + "learning_rate": 1.8284249868156378e-05, + "loss": 0.0153, + "step": 5500 + }, + { + "epoch": 9.229480737018426, + "grad_norm": 0.0667581558227539, + "learning_rate": 1.8243903694327992e-05, + "loss": 0.0139, + "step": 5510 + }, + { + "epoch": 9.246231155778894, + "grad_norm": 0.07833325862884521, + "learning_rate": 1.8203532885207272e-05, + "loss": 0.0156, + "step": 5520 + }, + { + "epoch": 9.262981574539364, + "grad_norm": 0.06736069172620773, + "learning_rate": 1.8163137747383675e-05, + "loss": 0.0127, + "step": 5530 + }, + { + "epoch": 9.279731993299832, + "grad_norm": 0.0944928526878357, + "learning_rate": 1.812271858763143e-05, + "loss": 0.0144, + "step": 5540 + }, + { + "epoch": 9.296482412060302, + "grad_norm": 0.09697055071592331, + "learning_rate": 1.8082275712907187e-05, + "loss": 0.0151, + "step": 5550 + }, + { + "epoch": 9.31323283082077, + "grad_norm": 0.09629473835229874, + "learning_rate": 1.8041809430347688e-05, + "loss": 0.0176, + "step": 5560 + }, + { + "epoch": 9.32998324958124, + "grad_norm": 0.07615918666124344, + "learning_rate": 1.8001320047267465e-05, + "loss": 0.0148, + "step": 5570 + }, + { + "epoch": 9.346733668341708, + "grad_norm": 0.07980474084615707, + "learning_rate": 1.796080787115646e-05, + "loss": 0.0149, + "step": 5580 + }, + { + "epoch": 9.363484087102178, + "grad_norm": 0.10057661682367325, + "learning_rate": 1.7920273209677722e-05, + "loss": 0.0167, + "step": 5590 + }, + { + "epoch": 9.380234505862646, + "grad_norm": 0.0847344771027565, + "learning_rate": 1.7879716370665068e-05, + "loss": 0.016, + "step": 5600 + }, + { + "epoch": 9.396984924623116, + "grad_norm": 0.07230829447507858, + "learning_rate": 1.783913766212073e-05, + "loss": 0.0161, + "step": 5610 + }, + { + "epoch": 9.413735343383584, + "grad_norm": 0.0874628946185112, + "learning_rate": 1.779853739221302e-05, + "loss": 0.0161, + "step": 5620 + }, + { + "epoch": 9.430485762144054, + "grad_norm": 0.07002632319927216, + "learning_rate": 1.775791586927401e-05, + "loss": 0.0148, + "step": 5630 + }, + { + "epoch": 9.447236180904522, + "grad_norm": 0.09408432990312576, + "learning_rate": 1.771727340179716e-05, + "loss": 0.013, + "step": 5640 + }, + { + "epoch": 9.463986599664992, + "grad_norm": 0.07826568186283112, + "learning_rate": 1.7676610298434993e-05, + "loss": 0.0155, + "step": 5650 + }, + { + "epoch": 9.48073701842546, + "grad_norm": 0.09074067324399948, + "learning_rate": 1.7635926867996758e-05, + "loss": 0.0144, + "step": 5660 + }, + { + "epoch": 9.49748743718593, + "grad_norm": 0.10472352802753448, + "learning_rate": 1.7595223419446064e-05, + "loss": 0.0159, + "step": 5670 + }, + { + "epoch": 9.514237855946398, + "grad_norm": 0.05583181977272034, + "learning_rate": 1.7554500261898545e-05, + "loss": 0.0138, + "step": 5680 + }, + { + "epoch": 9.530988274706868, + "grad_norm": 0.07133681327104568, + "learning_rate": 1.7513757704619513e-05, + "loss": 0.016, + "step": 5690 + }, + { + "epoch": 9.547738693467338, + "grad_norm": 0.057500433176755905, + "learning_rate": 1.747299605702161e-05, + "loss": 0.0135, + "step": 5700 + }, + { + "epoch": 9.564489112227806, + "grad_norm": 0.10911339521408081, + "learning_rate": 1.7432215628662463e-05, + "loss": 0.014, + "step": 5710 + }, + { + "epoch": 9.581239530988274, + "grad_norm": 0.09471974521875381, + "learning_rate": 1.7391416729242307e-05, + "loss": 0.0142, + "step": 5720 + }, + { + "epoch": 9.597989949748744, + "grad_norm": 0.08386453986167908, + "learning_rate": 1.7350599668601666e-05, + "loss": 0.0152, + "step": 5730 + }, + { + "epoch": 9.614740368509214, + "grad_norm": 0.09431836754083633, + "learning_rate": 1.730976475671899e-05, + "loss": 0.0181, + "step": 5740 + }, + { + "epoch": 9.631490787269682, + "grad_norm": 0.07204046100378036, + "learning_rate": 1.726891230370829e-05, + "loss": 0.0138, + "step": 5750 + }, + { + "epoch": 9.64824120603015, + "grad_norm": 0.08191115409135818, + "learning_rate": 1.7228042619816786e-05, + "loss": 0.0132, + "step": 5760 + }, + { + "epoch": 9.66499162479062, + "grad_norm": 0.06609786301851273, + "learning_rate": 1.7187156015422573e-05, + "loss": 0.0146, + "step": 5770 + }, + { + "epoch": 9.68174204355109, + "grad_norm": 0.08089634031057358, + "learning_rate": 1.7146252801032225e-05, + "loss": 0.014, + "step": 5780 + }, + { + "epoch": 9.698492462311558, + "grad_norm": 0.06636945903301239, + "learning_rate": 1.7105333287278465e-05, + "loss": 0.0126, + "step": 5790 + }, + { + "epoch": 9.715242881072026, + "grad_norm": 0.08253764361143112, + "learning_rate": 1.706439778491781e-05, + "loss": 0.0146, + "step": 5800 + }, + { + "epoch": 9.731993299832496, + "grad_norm": 0.06800912320613861, + "learning_rate": 1.702344660482818e-05, + "loss": 0.015, + "step": 5810 + }, + { + "epoch": 9.748743718592966, + "grad_norm": 0.05859517678618431, + "learning_rate": 1.6982480058006575e-05, + "loss": 0.0166, + "step": 5820 + }, + { + "epoch": 9.765494137353434, + "grad_norm": 0.09688255935907364, + "learning_rate": 1.6941498455566684e-05, + "loss": 0.0134, + "step": 5830 + }, + { + "epoch": 9.782244556113902, + "grad_norm": 0.07653312385082245, + "learning_rate": 1.690050210873654e-05, + "loss": 0.014, + "step": 5840 + }, + { + "epoch": 9.798994974874372, + "grad_norm": 0.08134609460830688, + "learning_rate": 1.6859491328856143e-05, + "loss": 0.0133, + "step": 5850 + }, + { + "epoch": 9.815745393634842, + "grad_norm": 0.06555905193090439, + "learning_rate": 1.6818466427375114e-05, + "loss": 0.0149, + "step": 5860 + }, + { + "epoch": 9.83249581239531, + "grad_norm": 0.06328080594539642, + "learning_rate": 1.67774277158503e-05, + "loss": 0.0134, + "step": 5870 + }, + { + "epoch": 9.849246231155778, + "grad_norm": 0.08666656166315079, + "learning_rate": 1.673637550594344e-05, + "loss": 0.0146, + "step": 5880 + }, + { + "epoch": 9.865996649916248, + "grad_norm": 0.06649037450551987, + "learning_rate": 1.669531010941879e-05, + "loss": 0.013, + "step": 5890 + }, + { + "epoch": 9.882747068676718, + "grad_norm": 0.0830390453338623, + "learning_rate": 1.6654231838140723e-05, + "loss": 0.013, + "step": 5900 + }, + { + "epoch": 9.899497487437186, + "grad_norm": 0.07208152860403061, + "learning_rate": 1.661314100407142e-05, + "loss": 0.0141, + "step": 5910 + }, + { + "epoch": 9.916247906197654, + "grad_norm": 0.07338222116231918, + "learning_rate": 1.657203791926845e-05, + "loss": 0.0123, + "step": 5920 + }, + { + "epoch": 9.932998324958124, + "grad_norm": 0.09653204679489136, + "learning_rate": 1.653092289588241e-05, + "loss": 0.0144, + "step": 5930 + }, + { + "epoch": 9.949748743718594, + "grad_norm": 0.07927566021680832, + "learning_rate": 1.6489796246154586e-05, + "loss": 0.0143, + "step": 5940 + }, + { + "epoch": 9.966499162479062, + "grad_norm": 0.057749975472688675, + "learning_rate": 1.6448658282414543e-05, + "loss": 0.0139, + "step": 5950 + }, + { + "epoch": 9.98324958123953, + "grad_norm": 0.07733950763940811, + "learning_rate": 1.6407509317077764e-05, + "loss": 0.0145, + "step": 5960 + }, + { + "epoch": 10.0, + "grad_norm": 0.06586837023496628, + "learning_rate": 1.6366349662643297e-05, + "loss": 0.0135, + "step": 5970 + }, + { + "epoch": 10.01675041876047, + "grad_norm": 0.09854725748300552, + "learning_rate": 1.6325179631691355e-05, + "loss": 0.0145, + "step": 5980 + }, + { + "epoch": 10.033500837520938, + "grad_norm": 0.08377289772033691, + "learning_rate": 1.6283999536880957e-05, + "loss": 0.0136, + "step": 5990 + }, + { + "epoch": 10.050251256281408, + "grad_norm": 0.07842150330543518, + "learning_rate": 1.6242809690947555e-05, + "loss": 0.0123, + "step": 6000 + }, + { + "epoch": 10.067001675041876, + "grad_norm": 0.08640950173139572, + "learning_rate": 1.6201610406700645e-05, + "loss": 0.0133, + "step": 6010 + }, + { + "epoch": 10.083752093802346, + "grad_norm": 0.1089024543762207, + "learning_rate": 1.6160401997021406e-05, + "loss": 0.014, + "step": 6020 + }, + { + "epoch": 10.100502512562814, + "grad_norm": 0.06064023822546005, + "learning_rate": 1.6119184774860322e-05, + "loss": 0.0146, + "step": 6030 + }, + { + "epoch": 10.117252931323284, + "grad_norm": 0.06450966000556946, + "learning_rate": 1.6077959053234796e-05, + "loss": 0.013, + "step": 6040 + }, + { + "epoch": 10.134003350083752, + "grad_norm": 0.07623566687107086, + "learning_rate": 1.6036725145226774e-05, + "loss": 0.0138, + "step": 6050 + }, + { + "epoch": 10.150753768844222, + "grad_norm": 0.0815720483660698, + "learning_rate": 1.5995483363980394e-05, + "loss": 0.0138, + "step": 6060 + }, + { + "epoch": 10.16750418760469, + "grad_norm": 0.061299752444028854, + "learning_rate": 1.595423402269956e-05, + "loss": 0.0137, + "step": 6070 + }, + { + "epoch": 10.18425460636516, + "grad_norm": 0.05670687556266785, + "learning_rate": 1.5912977434645608e-05, + "loss": 0.0165, + "step": 6080 + }, + { + "epoch": 10.201005025125628, + "grad_norm": 0.07310479134321213, + "learning_rate": 1.587171391313489e-05, + "loss": 0.0135, + "step": 6090 + }, + { + "epoch": 10.217755443886098, + "grad_norm": 0.06537134945392609, + "learning_rate": 1.5830443771536434e-05, + "loss": 0.0144, + "step": 6100 + }, + { + "epoch": 10.234505862646566, + "grad_norm": 0.07251140475273132, + "learning_rate": 1.578916732326953e-05, + "loss": 0.0157, + "step": 6110 + }, + { + "epoch": 10.251256281407034, + "grad_norm": 0.09199927002191544, + "learning_rate": 1.5747884881801364e-05, + "loss": 0.0144, + "step": 6120 + }, + { + "epoch": 10.268006700167504, + "grad_norm": 0.06975611299276352, + "learning_rate": 1.570659676064464e-05, + "loss": 0.0141, + "step": 6130 + }, + { + "epoch": 10.284757118927974, + "grad_norm": 0.0650688111782074, + "learning_rate": 1.5665303273355197e-05, + "loss": 0.0139, + "step": 6140 + }, + { + "epoch": 10.301507537688442, + "grad_norm": 0.11455053091049194, + "learning_rate": 1.562400473352962e-05, + "loss": 0.0134, + "step": 6150 + }, + { + "epoch": 10.318257956448912, + "grad_norm": 0.10255176573991776, + "learning_rate": 1.5582701454802867e-05, + "loss": 0.0126, + "step": 6160 + }, + { + "epoch": 10.33500837520938, + "grad_norm": 0.09504007548093796, + "learning_rate": 1.5541393750845886e-05, + "loss": 0.013, + "step": 6170 + }, + { + "epoch": 10.35175879396985, + "grad_norm": 0.053999241441488266, + "learning_rate": 1.5500081935363233e-05, + "loss": 0.0128, + "step": 6180 + }, + { + "epoch": 10.368509212730318, + "grad_norm": 0.07180154323577881, + "learning_rate": 1.5458766322090684e-05, + "loss": 0.0131, + "step": 6190 + }, + { + "epoch": 10.385259631490788, + "grad_norm": 0.07966731488704681, + "learning_rate": 1.5417447224792862e-05, + "loss": 0.0144, + "step": 6200 + }, + { + "epoch": 10.402010050251256, + "grad_norm": 0.08139584958553314, + "learning_rate": 1.5376124957260847e-05, + "loss": 0.0151, + "step": 6210 + }, + { + "epoch": 10.418760469011726, + "grad_norm": 0.06023263931274414, + "learning_rate": 1.5334799833309792e-05, + "loss": 0.0131, + "step": 6220 + }, + { + "epoch": 10.435510887772194, + "grad_norm": 0.09527294337749481, + "learning_rate": 1.529347216677654e-05, + "loss": 0.0134, + "step": 6230 + }, + { + "epoch": 10.452261306532664, + "grad_norm": 0.07697620987892151, + "learning_rate": 1.5252142271517259e-05, + "loss": 0.0139, + "step": 6240 + }, + { + "epoch": 10.469011725293132, + "grad_norm": 0.08630616217851639, + "learning_rate": 1.5210810461405019e-05, + "loss": 0.0127, + "step": 6250 + }, + { + "epoch": 10.485762144053602, + "grad_norm": 0.08324746787548065, + "learning_rate": 1.5169477050327457e-05, + "loss": 0.0137, + "step": 6260 + }, + { + "epoch": 10.50251256281407, + "grad_norm": 0.10764531046152115, + "learning_rate": 1.5128142352184346e-05, + "loss": 0.0116, + "step": 6270 + }, + { + "epoch": 10.519262981574538, + "grad_norm": 0.09146895259618759, + "learning_rate": 1.5086806680885245e-05, + "loss": 0.0125, + "step": 6280 + }, + { + "epoch": 10.536013400335008, + "grad_norm": 0.06823863834142685, + "learning_rate": 1.5045470350347106e-05, + "loss": 0.0116, + "step": 6290 + }, + { + "epoch": 10.552763819095478, + "grad_norm": 0.08591805398464203, + "learning_rate": 1.5004133674491876e-05, + "loss": 0.0125, + "step": 6300 + }, + { + "epoch": 10.569514237855946, + "grad_norm": 0.09628190100193024, + "learning_rate": 1.4962796967244137e-05, + "loss": 0.014, + "step": 6310 + }, + { + "epoch": 10.586264656616414, + "grad_norm": 0.09630268812179565, + "learning_rate": 1.4921460542528706e-05, + "loss": 0.0141, + "step": 6320 + }, + { + "epoch": 10.603015075376884, + "grad_norm": 0.06456310302019119, + "learning_rate": 1.4880124714268248e-05, + "loss": 0.0129, + "step": 6330 + }, + { + "epoch": 10.619765494137354, + "grad_norm": 0.06395621597766876, + "learning_rate": 1.4838789796380901e-05, + "loss": 0.0137, + "step": 6340 + }, + { + "epoch": 10.636515912897822, + "grad_norm": 0.06975158303976059, + "learning_rate": 1.4797456102777899e-05, + "loss": 0.0159, + "step": 6350 + }, + { + "epoch": 10.653266331658292, + "grad_norm": 0.07676661759614944, + "learning_rate": 1.4756123947361164e-05, + "loss": 0.0129, + "step": 6360 + }, + { + "epoch": 10.67001675041876, + "grad_norm": 0.07465475797653198, + "learning_rate": 1.4714793644020942e-05, + "loss": 0.013, + "step": 6370 + }, + { + "epoch": 10.68676716917923, + "grad_norm": 0.08438120037317276, + "learning_rate": 1.4673465506633417e-05, + "loss": 0.0148, + "step": 6380 + }, + { + "epoch": 10.703517587939698, + "grad_norm": 0.0682619959115982, + "learning_rate": 1.4632139849058328e-05, + "loss": 0.0132, + "step": 6390 + }, + { + "epoch": 10.720268006700168, + "grad_norm": 0.05983544513583183, + "learning_rate": 1.4590816985136563e-05, + "loss": 0.0153, + "step": 6400 + }, + { + "epoch": 10.737018425460636, + "grad_norm": 0.06861865520477295, + "learning_rate": 1.4549497228687812e-05, + "loss": 0.0141, + "step": 6410 + }, + { + "epoch": 10.753768844221106, + "grad_norm": 0.07951993495225906, + "learning_rate": 1.4508180893508157e-05, + "loss": 0.0145, + "step": 6420 + }, + { + "epoch": 10.770519262981574, + "grad_norm": 0.09159310907125473, + "learning_rate": 1.4466868293367708e-05, + "loss": 0.0128, + "step": 6430 + }, + { + "epoch": 10.787269681742044, + "grad_norm": 0.06889096647500992, + "learning_rate": 1.4425559742008196e-05, + "loss": 0.0144, + "step": 6440 + }, + { + "epoch": 10.804020100502512, + "grad_norm": 0.07343833148479462, + "learning_rate": 1.438425555314061e-05, + "loss": 0.0133, + "step": 6450 + }, + { + "epoch": 10.820770519262982, + "grad_norm": 0.08436884731054306, + "learning_rate": 1.4342956040442816e-05, + "loss": 0.0128, + "step": 6460 + }, + { + "epoch": 10.83752093802345, + "grad_norm": 0.0685647502541542, + "learning_rate": 1.4301661517557155e-05, + "loss": 0.0134, + "step": 6470 + }, + { + "epoch": 10.85427135678392, + "grad_norm": 0.07545416057109833, + "learning_rate": 1.426037229808808e-05, + "loss": 0.0125, + "step": 6480 + }, + { + "epoch": 10.871021775544389, + "grad_norm": 0.06606476753950119, + "learning_rate": 1.4219088695599773e-05, + "loss": 0.0131, + "step": 6490 + }, + { + "epoch": 10.887772194304858, + "grad_norm": 0.06588952988386154, + "learning_rate": 1.4177811023613749e-05, + "loss": 0.012, + "step": 6500 + }, + { + "epoch": 10.904522613065327, + "grad_norm": 0.0672588050365448, + "learning_rate": 1.4136539595606493e-05, + "loss": 0.0141, + "step": 6510 + }, + { + "epoch": 10.921273031825796, + "grad_norm": 0.06932391226291656, + "learning_rate": 1.4095274725007064e-05, + "loss": 0.0135, + "step": 6520 + }, + { + "epoch": 10.938023450586265, + "grad_norm": 0.055177588015794754, + "learning_rate": 1.4054016725194727e-05, + "loss": 0.0118, + "step": 6530 + }, + { + "epoch": 10.954773869346734, + "grad_norm": 0.06075577810406685, + "learning_rate": 1.4012765909496567e-05, + "loss": 0.0118, + "step": 6540 + }, + { + "epoch": 10.971524288107203, + "grad_norm": 0.08867711573839188, + "learning_rate": 1.3971522591185111e-05, + "loss": 0.0141, + "step": 6550 + }, + { + "epoch": 10.988274706867672, + "grad_norm": 0.06817186623811722, + "learning_rate": 1.3930287083475943e-05, + "loss": 0.0121, + "step": 6560 + }, + { + "epoch": 11.00502512562814, + "grad_norm": 0.06277189403772354, + "learning_rate": 1.3889059699525338e-05, + "loss": 0.0142, + "step": 6570 + }, + { + "epoch": 11.02177554438861, + "grad_norm": 0.05601593852043152, + "learning_rate": 1.3847840752427874e-05, + "loss": 0.0125, + "step": 6580 + }, + { + "epoch": 11.038525963149079, + "grad_norm": 0.14753024280071259, + "learning_rate": 1.380663055521405e-05, + "loss": 0.0133, + "step": 6590 + }, + { + "epoch": 11.055276381909549, + "grad_norm": 0.07654228061437607, + "learning_rate": 1.3765429420847928e-05, + "loss": 0.0126, + "step": 6600 + }, + { + "epoch": 11.072026800670017, + "grad_norm": 0.08025944232940674, + "learning_rate": 1.3724237662224738e-05, + "loss": 0.0146, + "step": 6610 + }, + { + "epoch": 11.088777219430487, + "grad_norm": 0.06721494346857071, + "learning_rate": 1.3683055592168497e-05, + "loss": 0.0159, + "step": 6620 + }, + { + "epoch": 11.105527638190955, + "grad_norm": 0.058076489716768265, + "learning_rate": 1.3641883523429665e-05, + "loss": 0.0145, + "step": 6630 + }, + { + "epoch": 11.122278056951425, + "grad_norm": 0.10511145740747452, + "learning_rate": 1.3600721768682729e-05, + "loss": 0.016, + "step": 6640 + }, + { + "epoch": 11.139028475711893, + "grad_norm": 0.07932860404253006, + "learning_rate": 1.3559570640523858e-05, + "loss": 0.0148, + "step": 6650 + }, + { + "epoch": 11.155778894472363, + "grad_norm": 0.1065361350774765, + "learning_rate": 1.351843045146852e-05, + "loss": 0.013, + "step": 6660 + }, + { + "epoch": 11.17252931323283, + "grad_norm": 0.09081228077411652, + "learning_rate": 1.3477301513949103e-05, + "loss": 0.0141, + "step": 6670 + }, + { + "epoch": 11.1892797319933, + "grad_norm": 0.06857980042695999, + "learning_rate": 1.3436184140312546e-05, + "loss": 0.0114, + "step": 6680 + }, + { + "epoch": 11.206030150753769, + "grad_norm": 0.06644202023744583, + "learning_rate": 1.3395078642817976e-05, + "loss": 0.0115, + "step": 6690 + }, + { + "epoch": 11.222780569514239, + "grad_norm": 0.07019989937543869, + "learning_rate": 1.3353985333634321e-05, + "loss": 0.0134, + "step": 6700 + }, + { + "epoch": 11.239530988274707, + "grad_norm": 0.06523793935775757, + "learning_rate": 1.3312904524837948e-05, + "loss": 0.0135, + "step": 6710 + }, + { + "epoch": 11.256281407035177, + "grad_norm": 0.08410218358039856, + "learning_rate": 1.3271836528410299e-05, + "loss": 0.0147, + "step": 6720 + }, + { + "epoch": 11.273031825795645, + "grad_norm": 0.10058104246854782, + "learning_rate": 1.3230781656235505e-05, + "loss": 0.015, + "step": 6730 + }, + { + "epoch": 11.289782244556115, + "grad_norm": 0.053105153143405914, + "learning_rate": 1.3189740220098026e-05, + "loss": 0.0118, + "step": 6740 + }, + { + "epoch": 11.306532663316583, + "grad_norm": 0.06775388866662979, + "learning_rate": 1.31487125316803e-05, + "loss": 0.0161, + "step": 6750 + }, + { + "epoch": 11.323283082077053, + "grad_norm": 0.06501705944538116, + "learning_rate": 1.3107698902560344e-05, + "loss": 0.0153, + "step": 6760 + }, + { + "epoch": 11.34003350083752, + "grad_norm": 0.08275182545185089, + "learning_rate": 1.3066699644209404e-05, + "loss": 0.0116, + "step": 6770 + }, + { + "epoch": 11.35678391959799, + "grad_norm": 0.06643733382225037, + "learning_rate": 1.3025715067989606e-05, + "loss": 0.0155, + "step": 6780 + }, + { + "epoch": 11.373534338358459, + "grad_norm": 0.08800774067640305, + "learning_rate": 1.2984745485151559e-05, + "loss": 0.0141, + "step": 6790 + }, + { + "epoch": 11.390284757118929, + "grad_norm": 0.10282844305038452, + "learning_rate": 1.2943791206832008e-05, + "loss": 0.0144, + "step": 6800 + }, + { + "epoch": 11.407035175879397, + "grad_norm": 0.08494739234447479, + "learning_rate": 1.290285254405148e-05, + "loss": 0.0138, + "step": 6810 + }, + { + "epoch": 11.423785594639867, + "grad_norm": 0.07839672267436981, + "learning_rate": 1.2861929807711901e-05, + "loss": 0.0148, + "step": 6820 + }, + { + "epoch": 11.440536013400335, + "grad_norm": 0.06659016013145447, + "learning_rate": 1.2821023308594256e-05, + "loss": 0.0144, + "step": 6830 + }, + { + "epoch": 11.457286432160805, + "grad_norm": 0.10885549336671829, + "learning_rate": 1.2780133357356218e-05, + "loss": 0.013, + "step": 6840 + }, + { + "epoch": 11.474036850921273, + "grad_norm": 0.07157129049301147, + "learning_rate": 1.273926026452978e-05, + "loss": 0.0138, + "step": 6850 + }, + { + "epoch": 11.490787269681743, + "grad_norm": 0.05953820422291756, + "learning_rate": 1.2698404340518922e-05, + "loss": 0.0129, + "step": 6860 + }, + { + "epoch": 11.507537688442211, + "grad_norm": 0.03799299895763397, + "learning_rate": 1.2657565895597222e-05, + "loss": 0.0106, + "step": 6870 + }, + { + "epoch": 11.52428810720268, + "grad_norm": 0.0774364024400711, + "learning_rate": 1.2616745239905527e-05, + "loss": 0.0138, + "step": 6880 + }, + { + "epoch": 11.541038525963149, + "grad_norm": 0.07412974536418915, + "learning_rate": 1.2575942683449585e-05, + "loss": 0.0134, + "step": 6890 + }, + { + "epoch": 11.557788944723619, + "grad_norm": 0.07800350338220596, + "learning_rate": 1.2535158536097691e-05, + "loss": 0.014, + "step": 6900 + }, + { + "epoch": 11.574539363484087, + "grad_norm": 0.08868993818759918, + "learning_rate": 1.249439310757833e-05, + "loss": 0.0139, + "step": 6910 + }, + { + "epoch": 11.591289782244557, + "grad_norm": 0.07635314017534256, + "learning_rate": 1.2453646707477843e-05, + "loss": 0.0134, + "step": 6920 + }, + { + "epoch": 11.608040201005025, + "grad_norm": 0.08172111213207245, + "learning_rate": 1.2412919645238049e-05, + "loss": 0.0124, + "step": 6930 + }, + { + "epoch": 11.624790619765495, + "grad_norm": 0.06785856932401657, + "learning_rate": 1.2372212230153911e-05, + "loss": 0.0118, + "step": 6940 + }, + { + "epoch": 11.641541038525963, + "grad_norm": 0.07923078536987305, + "learning_rate": 1.2331524771371197e-05, + "loss": 0.0131, + "step": 6950 + }, + { + "epoch": 11.658291457286433, + "grad_norm": 0.05836864933371544, + "learning_rate": 1.2290857577884109e-05, + "loss": 0.0126, + "step": 6960 + }, + { + "epoch": 11.675041876046901, + "grad_norm": 0.07780711352825165, + "learning_rate": 1.2250210958532943e-05, + "loss": 0.0141, + "step": 6970 + }, + { + "epoch": 11.691792294807371, + "grad_norm": 0.08186667412519455, + "learning_rate": 1.2209585222001762e-05, + "loss": 0.0129, + "step": 6980 + }, + { + "epoch": 11.708542713567839, + "grad_norm": 0.08738984167575836, + "learning_rate": 1.216898067681603e-05, + "loss": 0.0129, + "step": 6990 + }, + { + "epoch": 11.725293132328309, + "grad_norm": 0.07625941932201385, + "learning_rate": 1.2128397631340274e-05, + "loss": 0.0127, + "step": 7000 + }, + { + "epoch": 11.742043551088777, + "grad_norm": 0.06589437276124954, + "learning_rate": 1.2087836393775756e-05, + "loss": 0.0126, + "step": 7010 + }, + { + "epoch": 11.758793969849247, + "grad_norm": 0.06688161194324493, + "learning_rate": 1.2047297272158112e-05, + "loss": 0.0129, + "step": 7020 + }, + { + "epoch": 11.775544388609715, + "grad_norm": 0.08182154595851898, + "learning_rate": 1.2006780574355017e-05, + "loss": 0.0127, + "step": 7030 + }, + { + "epoch": 11.792294807370185, + "grad_norm": 0.06617710739374161, + "learning_rate": 1.1966286608063866e-05, + "loss": 0.0145, + "step": 7040 + }, + { + "epoch": 11.809045226130653, + "grad_norm": 0.060059934854507446, + "learning_rate": 1.1925815680809419e-05, + "loss": 0.0114, + "step": 7050 + }, + { + "epoch": 11.825795644891123, + "grad_norm": 0.08330146223306656, + "learning_rate": 1.1885368099941459e-05, + "loss": 0.0127, + "step": 7060 + }, + { + "epoch": 11.842546063651591, + "grad_norm": 0.08132729679346085, + "learning_rate": 1.1844944172632485e-05, + "loss": 0.0133, + "step": 7070 + }, + { + "epoch": 11.859296482412061, + "grad_norm": 0.08308562636375427, + "learning_rate": 1.1804544205875353e-05, + "loss": 0.0136, + "step": 7080 + }, + { + "epoch": 11.87604690117253, + "grad_norm": 0.07975098490715027, + "learning_rate": 1.1764168506480953e-05, + "loss": 0.0132, + "step": 7090 + }, + { + "epoch": 11.892797319932999, + "grad_norm": 0.07059597223997116, + "learning_rate": 1.1723817381075889e-05, + "loss": 0.0126, + "step": 7100 + }, + { + "epoch": 11.909547738693467, + "grad_norm": 0.07069030404090881, + "learning_rate": 1.1683491136100135e-05, + "loss": 0.0127, + "step": 7110 + }, + { + "epoch": 11.926298157453937, + "grad_norm": 0.06913162767887115, + "learning_rate": 1.1643190077804713e-05, + "loss": 0.0124, + "step": 7120 + }, + { + "epoch": 11.943048576214405, + "grad_norm": 0.07565716654062271, + "learning_rate": 1.1602914512249379e-05, + "loss": 0.0139, + "step": 7130 + }, + { + "epoch": 11.959798994974875, + "grad_norm": 0.05990506708621979, + "learning_rate": 1.156266474530028e-05, + "loss": 0.0111, + "step": 7140 + }, + { + "epoch": 11.976549413735343, + "grad_norm": 0.0784129723906517, + "learning_rate": 1.152244108262764e-05, + "loss": 0.0099, + "step": 7150 + }, + { + "epoch": 11.993299832495813, + "grad_norm": 0.0859949141740799, + "learning_rate": 1.1482243829703444e-05, + "loss": 0.0129, + "step": 7160 + }, + { + "epoch": 12.010050251256281, + "grad_norm": 0.0738619789481163, + "learning_rate": 1.1442073291799106e-05, + "loss": 0.0122, + "step": 7170 + }, + { + "epoch": 12.026800670016751, + "grad_norm": 0.09288010001182556, + "learning_rate": 1.1401929773983155e-05, + "loss": 0.0134, + "step": 7180 + }, + { + "epoch": 12.04355108877722, + "grad_norm": 0.07321544736623764, + "learning_rate": 1.1361813581118936e-05, + "loss": 0.0145, + "step": 7190 + }, + { + "epoch": 12.06030150753769, + "grad_norm": 0.07173691689968109, + "learning_rate": 1.1321725017862268e-05, + "loss": 0.0115, + "step": 7200 + }, + { + "epoch": 12.077051926298157, + "grad_norm": 0.058314502239227295, + "learning_rate": 1.1281664388659134e-05, + "loss": 0.0128, + "step": 7210 + }, + { + "epoch": 12.093802345058627, + "grad_norm": 0.07083640247583389, + "learning_rate": 1.1241631997743384e-05, + "loss": 0.0101, + "step": 7220 + }, + { + "epoch": 12.110552763819095, + "grad_norm": 0.065085269510746, + "learning_rate": 1.1201628149134422e-05, + "loss": 0.0137, + "step": 7230 + }, + { + "epoch": 12.127303182579565, + "grad_norm": 0.0781814306974411, + "learning_rate": 1.116165314663489e-05, + "loss": 0.0146, + "step": 7240 + }, + { + "epoch": 12.144053601340033, + "grad_norm": 0.09068423509597778, + "learning_rate": 1.1121707293828353e-05, + "loss": 0.014, + "step": 7250 + }, + { + "epoch": 12.160804020100503, + "grad_norm": 0.08548370748758316, + "learning_rate": 1.1081790894077022e-05, + "loss": 0.0121, + "step": 7260 + }, + { + "epoch": 12.177554438860971, + "grad_norm": 0.07648133486509323, + "learning_rate": 1.104190425051941e-05, + "loss": 0.0131, + "step": 7270 + }, + { + "epoch": 12.194304857621441, + "grad_norm": 0.09778308868408203, + "learning_rate": 1.1002047666068066e-05, + "loss": 0.0139, + "step": 7280 + }, + { + "epoch": 12.21105527638191, + "grad_norm": 0.07223774492740631, + "learning_rate": 1.0962221443407257e-05, + "loss": 0.0137, + "step": 7290 + }, + { + "epoch": 12.22780569514238, + "grad_norm": 0.0755016952753067, + "learning_rate": 1.0922425884990673e-05, + "loss": 0.0119, + "step": 7300 + }, + { + "epoch": 12.244556113902847, + "grad_norm": 0.0679793432354927, + "learning_rate": 1.0882661293039124e-05, + "loss": 0.0127, + "step": 7310 + }, + { + "epoch": 12.261306532663317, + "grad_norm": 0.08829930424690247, + "learning_rate": 1.0842927969538258e-05, + "loss": 0.0141, + "step": 7320 + }, + { + "epoch": 12.278056951423785, + "grad_norm": 0.06518194079399109, + "learning_rate": 1.0803226216236256e-05, + "loss": 0.0125, + "step": 7330 + }, + { + "epoch": 12.294807370184255, + "grad_norm": 0.08196015655994415, + "learning_rate": 1.076355633464154e-05, + "loss": 0.0114, + "step": 7340 + }, + { + "epoch": 12.311557788944723, + "grad_norm": 0.06805971264839172, + "learning_rate": 1.0723918626020498e-05, + "loss": 0.0118, + "step": 7350 + }, + { + "epoch": 12.328308207705193, + "grad_norm": 0.0839708149433136, + "learning_rate": 1.0684313391395176e-05, + "loss": 0.0148, + "step": 7360 + }, + { + "epoch": 12.345058626465661, + "grad_norm": 0.07113301753997803, + "learning_rate": 1.0644740931541005e-05, + "loss": 0.015, + "step": 7370 + }, + { + "epoch": 12.361809045226131, + "grad_norm": 0.08832506835460663, + "learning_rate": 1.0605201546984517e-05, + "loss": 0.0148, + "step": 7380 + }, + { + "epoch": 12.3785594639866, + "grad_norm": 0.09089552611112595, + "learning_rate": 1.0565695538001056e-05, + "loss": 0.016, + "step": 7390 + }, + { + "epoch": 12.39530988274707, + "grad_norm": 0.06780517846345901, + "learning_rate": 1.0526223204612495e-05, + "loss": 0.0119, + "step": 7400 + }, + { + "epoch": 12.412060301507537, + "grad_norm": 0.07529979944229126, + "learning_rate": 1.0486784846584974e-05, + "loss": 0.0111, + "step": 7410 + }, + { + "epoch": 12.428810720268007, + "grad_norm": 0.08596140146255493, + "learning_rate": 1.0447380763426607e-05, + "loss": 0.0122, + "step": 7420 + }, + { + "epoch": 12.445561139028475, + "grad_norm": 0.06264451146125793, + "learning_rate": 1.040801125438521e-05, + "loss": 0.013, + "step": 7430 + }, + { + "epoch": 12.462311557788945, + "grad_norm": 0.06601212918758392, + "learning_rate": 1.0368676618446042e-05, + "loss": 0.0132, + "step": 7440 + }, + { + "epoch": 12.479061976549414, + "grad_norm": 0.08287645131349564, + "learning_rate": 1.0329377154329512e-05, + "loss": 0.0127, + "step": 7450 + }, + { + "epoch": 12.495812395309883, + "grad_norm": 0.10921191424131393, + "learning_rate": 1.0290113160488925e-05, + "loss": 0.0125, + "step": 7460 + }, + { + "epoch": 12.512562814070352, + "grad_norm": 0.07738841325044632, + "learning_rate": 1.0250884935108223e-05, + "loss": 0.0142, + "step": 7470 + }, + { + "epoch": 12.529313232830821, + "grad_norm": 0.07163051515817642, + "learning_rate": 1.0211692776099698e-05, + "loss": 0.0118, + "step": 7480 + }, + { + "epoch": 12.54606365159129, + "grad_norm": 0.07590950280427933, + "learning_rate": 1.0172536981101747e-05, + "loss": 0.0093, + "step": 7490 + }, + { + "epoch": 12.56281407035176, + "grad_norm": 0.08660171180963516, + "learning_rate": 1.0133417847476606e-05, + "loss": 0.012, + "step": 7500 + }, + { + "epoch": 12.579564489112228, + "grad_norm": 0.07874178141355515, + "learning_rate": 1.0094335672308095e-05, + "loss": 0.0117, + "step": 7510 + }, + { + "epoch": 12.596314907872697, + "grad_norm": 0.07014073431491852, + "learning_rate": 1.0055290752399353e-05, + "loss": 0.0138, + "step": 7520 + }, + { + "epoch": 12.613065326633166, + "grad_norm": 0.07678911834955215, + "learning_rate": 1.0016283384270594e-05, + "loss": 0.0117, + "step": 7530 + }, + { + "epoch": 12.629815745393635, + "grad_norm": 0.06955492496490479, + "learning_rate": 9.977313864156854e-06, + "loss": 0.0108, + "step": 7540 + }, + { + "epoch": 12.646566164154104, + "grad_norm": 0.08744408190250397, + "learning_rate": 9.938382488005729e-06, + "loss": 0.0116, + "step": 7550 + }, + { + "epoch": 12.663316582914574, + "grad_norm": 0.08456790447235107, + "learning_rate": 9.89948955147515e-06, + "loss": 0.0142, + "step": 7560 + }, + { + "epoch": 12.680067001675042, + "grad_norm": 0.08989357948303223, + "learning_rate": 9.860635349931116e-06, + "loss": 0.011, + "step": 7570 + }, + { + "epoch": 12.696817420435512, + "grad_norm": 0.05640851706266403, + "learning_rate": 9.821820178445455e-06, + "loss": 0.0116, + "step": 7580 + }, + { + "epoch": 12.71356783919598, + "grad_norm": 0.07872806489467621, + "learning_rate": 9.783044331793605e-06, + "loss": 0.0106, + "step": 7590 + }, + { + "epoch": 12.73031825795645, + "grad_norm": 0.08626449853181839, + "learning_rate": 9.744308104452338e-06, + "loss": 0.0118, + "step": 7600 + }, + { + "epoch": 12.747068676716918, + "grad_norm": 0.05731945484876633, + "learning_rate": 9.705611790597562e-06, + "loss": 0.0114, + "step": 7610 + }, + { + "epoch": 12.763819095477388, + "grad_norm": 0.06404105573892593, + "learning_rate": 9.666955684102043e-06, + "loss": 0.0123, + "step": 7620 + }, + { + "epoch": 12.780569514237856, + "grad_norm": 0.0529644712805748, + "learning_rate": 9.62834007853323e-06, + "loss": 0.0127, + "step": 7630 + }, + { + "epoch": 12.797319932998326, + "grad_norm": 0.10183175653219223, + "learning_rate": 9.589765267150975e-06, + "loss": 0.0123, + "step": 7640 + }, + { + "epoch": 12.814070351758794, + "grad_norm": 0.061799515038728714, + "learning_rate": 9.551231542905326e-06, + "loss": 0.0134, + "step": 7650 + }, + { + "epoch": 12.830820770519264, + "grad_norm": 0.04345851019024849, + "learning_rate": 9.512739198434312e-06, + "loss": 0.0137, + "step": 7660 + }, + { + "epoch": 12.847571189279732, + "grad_norm": 0.057996947318315506, + "learning_rate": 9.474288526061698e-06, + "loss": 0.0144, + "step": 7670 + }, + { + "epoch": 12.864321608040202, + "grad_norm": 0.07847338914871216, + "learning_rate": 9.435879817794784e-06, + "loss": 0.0123, + "step": 7680 + }, + { + "epoch": 12.88107202680067, + "grad_norm": 0.0713241770863533, + "learning_rate": 9.39751336532218e-06, + "loss": 0.0101, + "step": 7690 + }, + { + "epoch": 12.89782244556114, + "grad_norm": 0.07546516507863998, + "learning_rate": 9.359189460011591e-06, + "loss": 0.0139, + "step": 7700 + }, + { + "epoch": 12.914572864321608, + "grad_norm": 0.04970499873161316, + "learning_rate": 9.3209083929076e-06, + "loss": 0.0142, + "step": 7710 + }, + { + "epoch": 12.931323283082078, + "grad_norm": 0.08262616395950317, + "learning_rate": 9.282670454729471e-06, + "loss": 0.0115, + "step": 7720 + }, + { + "epoch": 12.948073701842546, + "grad_norm": 0.055856745690107346, + "learning_rate": 9.244475935868927e-06, + "loss": 0.0117, + "step": 7730 + }, + { + "epoch": 12.964824120603016, + "grad_norm": 0.04862665757536888, + "learning_rate": 9.206325126387945e-06, + "loss": 0.0104, + "step": 7740 + }, + { + "epoch": 12.981574539363484, + "grad_norm": 0.06921498477458954, + "learning_rate": 9.168218316016572e-06, + "loss": 0.0123, + "step": 7750 + }, + { + "epoch": 12.998324958123954, + "grad_norm": 0.09353175759315491, + "learning_rate": 9.130155794150698e-06, + "loss": 0.0121, + "step": 7760 + }, + { + "epoch": 13.015075376884422, + "grad_norm": 0.07180416584014893, + "learning_rate": 9.092137849849876e-06, + "loss": 0.0133, + "step": 7770 + }, + { + "epoch": 13.031825795644892, + "grad_norm": 0.07870851457118988, + "learning_rate": 9.054164771835128e-06, + "loss": 0.0123, + "step": 7780 + }, + { + "epoch": 13.04857621440536, + "grad_norm": 0.06702952831983566, + "learning_rate": 9.016236848486736e-06, + "loss": 0.0107, + "step": 7790 + }, + { + "epoch": 13.06532663316583, + "grad_norm": 0.07762094587087631, + "learning_rate": 8.978354367842068e-06, + "loss": 0.011, + "step": 7800 + }, + { + "epoch": 13.082077051926298, + "grad_norm": 0.07945635914802551, + "learning_rate": 8.940517617593384e-06, + "loss": 0.0119, + "step": 7810 + }, + { + "epoch": 13.098827470686768, + "grad_norm": 0.06439702212810516, + "learning_rate": 8.902726885085655e-06, + "loss": 0.0095, + "step": 7820 + }, + { + "epoch": 13.115577889447236, + "grad_norm": 0.0856538638472557, + "learning_rate": 8.864982457314365e-06, + "loss": 0.0142, + "step": 7830 + }, + { + "epoch": 13.132328308207706, + "grad_norm": 0.07062462717294693, + "learning_rate": 8.827284620923369e-06, + "loss": 0.0126, + "step": 7840 + }, + { + "epoch": 13.149078726968174, + "grad_norm": 0.08214748650789261, + "learning_rate": 8.789633662202667e-06, + "loss": 0.0122, + "step": 7850 + }, + { + "epoch": 13.165829145728644, + "grad_norm": 0.06878134608268738, + "learning_rate": 8.752029867086264e-06, + "loss": 0.0119, + "step": 7860 + }, + { + "epoch": 13.182579564489112, + "grad_norm": 0.10038811713457108, + "learning_rate": 8.714473521149994e-06, + "loss": 0.012, + "step": 7870 + }, + { + "epoch": 13.199329983249582, + "grad_norm": 0.05590192601084709, + "learning_rate": 8.676964909609334e-06, + "loss": 0.0121, + "step": 7880 + }, + { + "epoch": 13.21608040201005, + "grad_norm": 0.08674315363168716, + "learning_rate": 8.639504317317258e-06, + "loss": 0.013, + "step": 7890 + }, + { + "epoch": 13.23283082077052, + "grad_norm": 0.07082588225603104, + "learning_rate": 8.602092028762073e-06, + "loss": 0.0133, + "step": 7900 + }, + { + "epoch": 13.249581239530988, + "grad_norm": 0.09573926776647568, + "learning_rate": 8.56472832806523e-06, + "loss": 0.0137, + "step": 7910 + }, + { + "epoch": 13.266331658291458, + "grad_norm": 0.073761947453022, + "learning_rate": 8.527413498979204e-06, + "loss": 0.0112, + "step": 7920 + }, + { + "epoch": 13.283082077051926, + "grad_norm": 0.06695816665887833, + "learning_rate": 8.490147824885328e-06, + "loss": 0.0112, + "step": 7930 + }, + { + "epoch": 13.299832495812396, + "grad_norm": 0.09559398889541626, + "learning_rate": 8.452931588791614e-06, + "loss": 0.0126, + "step": 7940 + }, + { + "epoch": 13.316582914572864, + "grad_norm": 0.0765177384018898, + "learning_rate": 8.415765073330642e-06, + "loss": 0.0121, + "step": 7950 + }, + { + "epoch": 13.333333333333334, + "grad_norm": 0.06615452468395233, + "learning_rate": 8.378648560757396e-06, + "loss": 0.0131, + "step": 7960 + }, + { + "epoch": 13.350083752093802, + "grad_norm": 0.08720266819000244, + "learning_rate": 8.34158233294711e-06, + "loss": 0.0135, + "step": 7970 + }, + { + "epoch": 13.366834170854272, + "grad_norm": 0.07875142246484756, + "learning_rate": 8.304566671393151e-06, + "loss": 0.0126, + "step": 7980 + }, + { + "epoch": 13.38358458961474, + "grad_norm": 0.04822768643498421, + "learning_rate": 8.267601857204869e-06, + "loss": 0.0119, + "step": 7990 + }, + { + "epoch": 13.40033500837521, + "grad_norm": 0.062463391572237015, + "learning_rate": 8.230688171105446e-06, + "loss": 0.011, + "step": 8000 + }, + { + "epoch": 13.417085427135678, + "grad_norm": 0.06186162680387497, + "learning_rate": 8.193825893429796e-06, + "loss": 0.0108, + "step": 8010 + }, + { + "epoch": 13.433835845896148, + "grad_norm": 0.06208765134215355, + "learning_rate": 8.157015304122424e-06, + "loss": 0.011, + "step": 8020 + }, + { + "epoch": 13.450586264656616, + "grad_norm": 0.06548567116260529, + "learning_rate": 8.120256682735282e-06, + "loss": 0.0111, + "step": 8030 + }, + { + "epoch": 13.467336683417086, + "grad_norm": 0.09333834797143936, + "learning_rate": 8.083550308425658e-06, + "loss": 0.0118, + "step": 8040 + }, + { + "epoch": 13.484087102177554, + "grad_norm": 0.08424598723649979, + "learning_rate": 8.04689645995407e-06, + "loss": 0.0115, + "step": 8050 + }, + { + "epoch": 13.500837520938024, + "grad_norm": 0.08424214273691177, + "learning_rate": 8.010295415682144e-06, + "loss": 0.0113, + "step": 8060 + }, + { + "epoch": 13.517587939698492, + "grad_norm": 0.05681629851460457, + "learning_rate": 7.97374745357047e-06, + "loss": 0.012, + "step": 8070 + }, + { + "epoch": 13.534338358458962, + "grad_norm": 0.09238583594560623, + "learning_rate": 7.937252851176529e-06, + "loss": 0.0124, + "step": 8080 + }, + { + "epoch": 13.55108877721943, + "grad_norm": 0.06826833635568619, + "learning_rate": 7.900811885652577e-06, + "loss": 0.0124, + "step": 8090 + }, + { + "epoch": 13.5678391959799, + "grad_norm": 0.06261388212442398, + "learning_rate": 7.864424833743515e-06, + "loss": 0.0114, + "step": 8100 + }, + { + "epoch": 13.584589614740368, + "grad_norm": 0.056088827550411224, + "learning_rate": 7.828091971784816e-06, + "loss": 0.0111, + "step": 8110 + }, + { + "epoch": 13.601340033500838, + "grad_norm": 0.09993434697389603, + "learning_rate": 7.791813575700425e-06, + "loss": 0.012, + "step": 8120 + }, + { + "epoch": 13.618090452261306, + "grad_norm": 0.04731757566332817, + "learning_rate": 7.755589921000634e-06, + "loss": 0.0128, + "step": 8130 + }, + { + "epoch": 13.634840871021776, + "grad_norm": 0.0865129604935646, + "learning_rate": 7.71942128278003e-06, + "loss": 0.0112, + "step": 8140 + }, + { + "epoch": 13.651591289782244, + "grad_norm": 0.07481952011585236, + "learning_rate": 7.683307935715383e-06, + "loss": 0.0124, + "step": 8150 + }, + { + "epoch": 13.668341708542714, + "grad_norm": 0.07770776748657227, + "learning_rate": 7.647250154063548e-06, + "loss": 0.0132, + "step": 8160 + }, + { + "epoch": 13.685092127303182, + "grad_norm": 0.08305969834327698, + "learning_rate": 7.611248211659417e-06, + "loss": 0.0129, + "step": 8170 + }, + { + "epoch": 13.701842546063652, + "grad_norm": 0.0771985650062561, + "learning_rate": 7.575302381913813e-06, + "loss": 0.0101, + "step": 8180 + }, + { + "epoch": 13.71859296482412, + "grad_norm": 0.06562207639217377, + "learning_rate": 7.539412937811414e-06, + "loss": 0.0116, + "step": 8190 + }, + { + "epoch": 13.73534338358459, + "grad_norm": 0.05711517855525017, + "learning_rate": 7.503580151908694e-06, + "loss": 0.0126, + "step": 8200 + }, + { + "epoch": 13.752093802345058, + "grad_norm": 0.05801519751548767, + "learning_rate": 7.467804296331849e-06, + "loss": 0.0109, + "step": 8210 + }, + { + "epoch": 13.768844221105528, + "grad_norm": 0.08770089596509933, + "learning_rate": 7.432085642774713e-06, + "loss": 0.0109, + "step": 8220 + }, + { + "epoch": 13.785594639865996, + "grad_norm": 0.05994917079806328, + "learning_rate": 7.396424462496722e-06, + "loss": 0.0099, + "step": 8230 + }, + { + "epoch": 13.802345058626466, + "grad_norm": 0.08041512221097946, + "learning_rate": 7.360821026320844e-06, + "loss": 0.0121, + "step": 8240 + }, + { + "epoch": 13.819095477386934, + "grad_norm": 0.0712173655629158, + "learning_rate": 7.325275604631496e-06, + "loss": 0.0127, + "step": 8250 + }, + { + "epoch": 13.835845896147404, + "grad_norm": 0.07469300925731659, + "learning_rate": 7.289788467372536e-06, + "loss": 0.0111, + "step": 8260 + }, + { + "epoch": 13.852596314907872, + "grad_norm": 0.07342658936977386, + "learning_rate": 7.25435988404519e-06, + "loss": 0.0109, + "step": 8270 + }, + { + "epoch": 13.869346733668342, + "grad_norm": 0.08993782103061676, + "learning_rate": 7.218990123705987e-06, + "loss": 0.0123, + "step": 8280 + }, + { + "epoch": 13.88609715242881, + "grad_norm": 0.06399506330490112, + "learning_rate": 7.183679454964752e-06, + "loss": 0.0111, + "step": 8290 + }, + { + "epoch": 13.90284757118928, + "grad_norm": 0.08531822264194489, + "learning_rate": 7.148428145982548e-06, + "loss": 0.0113, + "step": 8300 + }, + { + "epoch": 13.919597989949748, + "grad_norm": 0.06432479619979858, + "learning_rate": 7.113236464469625e-06, + "loss": 0.0113, + "step": 8310 + }, + { + "epoch": 13.936348408710218, + "grad_norm": 0.06948276609182358, + "learning_rate": 7.078104677683417e-06, + "loss": 0.0127, + "step": 8320 + }, + { + "epoch": 13.953098827470686, + "grad_norm": 0.07048949599266052, + "learning_rate": 7.043033052426498e-06, + "loss": 0.0124, + "step": 8330 + }, + { + "epoch": 13.969849246231156, + "grad_norm": 0.11620538681745529, + "learning_rate": 7.00802185504454e-06, + "loss": 0.0119, + "step": 8340 + }, + { + "epoch": 13.986599664991624, + "grad_norm": 0.06149009242653847, + "learning_rate": 6.973071351424319e-06, + "loss": 0.0107, + "step": 8350 + }, + { + "epoch": 14.003350083752094, + "grad_norm": 0.061725832521915436, + "learning_rate": 6.9381818069916835e-06, + "loss": 0.0105, + "step": 8360 + }, + { + "epoch": 14.020100502512562, + "grad_norm": 0.09621080011129379, + "learning_rate": 6.903353486709525e-06, + "loss": 0.0123, + "step": 8370 + }, + { + "epoch": 14.036850921273032, + "grad_norm": 0.05495775490999222, + "learning_rate": 6.868586655075787e-06, + "loss": 0.0091, + "step": 8380 + }, + { + "epoch": 14.0536013400335, + "grad_norm": 0.06942631304264069, + "learning_rate": 6.833881576121452e-06, + "loss": 0.0133, + "step": 8390 + }, + { + "epoch": 14.07035175879397, + "grad_norm": 0.07615401595830917, + "learning_rate": 6.7992385134085175e-06, + "loss": 0.0108, + "step": 8400 + }, + { + "epoch": 14.087102177554438, + "grad_norm": 0.08464915305376053, + "learning_rate": 6.764657730028022e-06, + "loss": 0.0137, + "step": 8410 + }, + { + "epoch": 14.103852596314908, + "grad_norm": 0.07944124191999435, + "learning_rate": 6.7301394885980335e-06, + "loss": 0.0112, + "step": 8420 + }, + { + "epoch": 14.120603015075377, + "grad_norm": 0.07688767462968826, + "learning_rate": 6.69568405126165e-06, + "loss": 0.0106, + "step": 8430 + }, + { + "epoch": 14.137353433835846, + "grad_norm": 0.06504269689321518, + "learning_rate": 6.661291679685004e-06, + "loss": 0.0105, + "step": 8440 + }, + { + "epoch": 14.154103852596315, + "grad_norm": 0.06587643921375275, + "learning_rate": 6.626962635055308e-06, + "loss": 0.0134, + "step": 8450 + }, + { + "epoch": 14.170854271356784, + "grad_norm": 0.07884813100099564, + "learning_rate": 6.5926971780788394e-06, + "loss": 0.0132, + "step": 8460 + }, + { + "epoch": 14.187604690117253, + "grad_norm": 0.04981286823749542, + "learning_rate": 6.558495568978959e-06, + "loss": 0.0109, + "step": 8470 + }, + { + "epoch": 14.204355108877722, + "grad_norm": 0.08177836984395981, + "learning_rate": 6.5243580674941575e-06, + "loss": 0.0104, + "step": 8480 + }, + { + "epoch": 14.22110552763819, + "grad_norm": 0.07541631907224655, + "learning_rate": 6.490284932876075e-06, + "loss": 0.0123, + "step": 8490 + }, + { + "epoch": 14.23785594639866, + "grad_norm": 0.06907821446657181, + "learning_rate": 6.4562764238875085e-06, + "loss": 0.0113, + "step": 8500 + }, + { + "epoch": 14.254606365159129, + "grad_norm": 0.059419356286525726, + "learning_rate": 6.4223327988004835e-06, + "loss": 0.0116, + "step": 8510 + }, + { + "epoch": 14.271356783919598, + "grad_norm": 0.062080204486846924, + "learning_rate": 6.388454315394275e-06, + "loss": 0.0102, + "step": 8520 + }, + { + "epoch": 14.288107202680067, + "grad_norm": 0.07892616838216782, + "learning_rate": 6.3546412309534355e-06, + "loss": 0.0106, + "step": 8530 + }, + { + "epoch": 14.304857621440537, + "grad_norm": 0.08079873770475388, + "learning_rate": 6.32089380226587e-06, + "loss": 0.0131, + "step": 8540 + }, + { + "epoch": 14.321608040201005, + "grad_norm": 0.06901709735393524, + "learning_rate": 6.287212285620876e-06, + "loss": 0.0113, + "step": 8550 + }, + { + "epoch": 14.338358458961475, + "grad_norm": 0.07588738203048706, + "learning_rate": 6.25359693680717e-06, + "loss": 0.012, + "step": 8560 + }, + { + "epoch": 14.355108877721943, + "grad_norm": 0.08524017035961151, + "learning_rate": 6.22004801111099e-06, + "loss": 0.015, + "step": 8570 + }, + { + "epoch": 14.371859296482413, + "grad_norm": 0.0674111470580101, + "learning_rate": 6.186565763314131e-06, + "loss": 0.0111, + "step": 8580 + }, + { + "epoch": 14.38860971524288, + "grad_norm": 0.0773531049489975, + "learning_rate": 6.153150447691996e-06, + "loss": 0.0107, + "step": 8590 + }, + { + "epoch": 14.40536013400335, + "grad_norm": 0.05608673393726349, + "learning_rate": 6.119802318011707e-06, + "loss": 0.0115, + "step": 8600 + }, + { + "epoch": 14.422110552763819, + "grad_norm": 0.06338393688201904, + "learning_rate": 6.0865216275301425e-06, + "loss": 0.0122, + "step": 8610 + }, + { + "epoch": 14.438860971524289, + "grad_norm": 0.074888676404953, + "learning_rate": 6.0533086289920176e-06, + "loss": 0.0106, + "step": 8620 + }, + { + "epoch": 14.455611390284757, + "grad_norm": 0.08034703880548477, + "learning_rate": 6.020163574627984e-06, + "loss": 0.0133, + "step": 8630 + }, + { + "epoch": 14.472361809045227, + "grad_norm": 0.0592302568256855, + "learning_rate": 5.987086716152702e-06, + "loss": 0.011, + "step": 8640 + }, + { + "epoch": 14.489112227805695, + "grad_norm": 0.0759633481502533, + "learning_rate": 5.954078304762916e-06, + "loss": 0.012, + "step": 8650 + }, + { + "epoch": 14.505862646566165, + "grad_norm": 0.0660175308585167, + "learning_rate": 5.921138591135575e-06, + "loss": 0.012, + "step": 8660 + }, + { + "epoch": 14.522613065326633, + "grad_norm": 0.08975532650947571, + "learning_rate": 5.888267825425917e-06, + "loss": 0.0121, + "step": 8670 + }, + { + "epoch": 14.539363484087103, + "grad_norm": 0.07022812962532043, + "learning_rate": 5.8554662572655414e-06, + "loss": 0.0131, + "step": 8680 + }, + { + "epoch": 14.55611390284757, + "grad_norm": 0.057023871690034866, + "learning_rate": 5.822734135760566e-06, + "loss": 0.0119, + "step": 8690 + }, + { + "epoch": 14.57286432160804, + "grad_norm": 0.05912158265709877, + "learning_rate": 5.790071709489698e-06, + "loss": 0.0096, + "step": 8700 + }, + { + "epoch": 14.589614740368509, + "grad_norm": 0.07548556476831436, + "learning_rate": 5.757479226502347e-06, + "loss": 0.0117, + "step": 8710 + }, + { + "epoch": 14.606365159128979, + "grad_norm": 0.07981330156326294, + "learning_rate": 5.724956934316762e-06, + "loss": 0.0132, + "step": 8720 + }, + { + "epoch": 14.623115577889447, + "grad_norm": 0.07118304818868637, + "learning_rate": 5.692505079918138e-06, + "loss": 0.0113, + "step": 8730 + }, + { + "epoch": 14.639865996649917, + "grad_norm": 0.07005375623703003, + "learning_rate": 5.660123909756726e-06, + "loss": 0.011, + "step": 8740 + }, + { + "epoch": 14.656616415410385, + "grad_norm": 0.04754802584648132, + "learning_rate": 5.627813669746001e-06, + "loss": 0.012, + "step": 8750 + }, + { + "epoch": 14.673366834170855, + "grad_norm": 0.09637465327978134, + "learning_rate": 5.595574605260758e-06, + "loss": 0.0103, + "step": 8760 + }, + { + "epoch": 14.690117252931323, + "grad_norm": 0.09745600819587708, + "learning_rate": 5.563406961135256e-06, + "loss": 0.0129, + "step": 8770 + }, + { + "epoch": 14.706867671691793, + "grad_norm": 0.07260012626647949, + "learning_rate": 5.531310981661374e-06, + "loss": 0.0121, + "step": 8780 + }, + { + "epoch": 14.72361809045226, + "grad_norm": 0.07536528259515762, + "learning_rate": 5.499286910586749e-06, + "loss": 0.0118, + "step": 8790 + }, + { + "epoch": 14.74036850921273, + "grad_norm": 0.10453300923109055, + "learning_rate": 5.467334991112902e-06, + "loss": 0.012, + "step": 8800 + }, + { + "epoch": 14.757118927973199, + "grad_norm": 0.06679607182741165, + "learning_rate": 5.435455465893427e-06, + "loss": 0.01, + "step": 8810 + }, + { + "epoch": 14.773869346733669, + "grad_norm": 0.07301633059978485, + "learning_rate": 5.403648577032135e-06, + "loss": 0.0119, + "step": 8820 + }, + { + "epoch": 14.790619765494137, + "grad_norm": 0.08366880565881729, + "learning_rate": 5.3719145660812e-06, + "loss": 0.0115, + "step": 8830 + }, + { + "epoch": 14.807370184254607, + "grad_norm": 0.06086117401719093, + "learning_rate": 5.340253674039336e-06, + "loss": 0.0131, + "step": 8840 + }, + { + "epoch": 14.824120603015075, + "grad_norm": 0.08130698651075363, + "learning_rate": 5.308666141349976e-06, + "loss": 0.0114, + "step": 8850 + }, + { + "epoch": 14.840871021775545, + "grad_norm": 0.06073309853672981, + "learning_rate": 5.277152207899443e-06, + "loss": 0.0113, + "step": 8860 + }, + { + "epoch": 14.857621440536013, + "grad_norm": 0.08015117794275284, + "learning_rate": 5.245712113015101e-06, + "loss": 0.0126, + "step": 8870 + }, + { + "epoch": 14.874371859296483, + "grad_norm": 0.0700366273522377, + "learning_rate": 5.2143460954635805e-06, + "loss": 0.0113, + "step": 8880 + }, + { + "epoch": 14.891122278056951, + "grad_norm": 0.07298661768436432, + "learning_rate": 5.183054393448937e-06, + "loss": 0.0126, + "step": 8890 + }, + { + "epoch": 14.90787269681742, + "grad_norm": 0.055050741881132126, + "learning_rate": 5.151837244610842e-06, + "loss": 0.0107, + "step": 8900 + }, + { + "epoch": 14.924623115577889, + "grad_norm": 0.09468694031238556, + "learning_rate": 5.120694886022792e-06, + "loss": 0.0127, + "step": 8910 + }, + { + "epoch": 14.941373534338359, + "grad_norm": 0.06892071664333344, + "learning_rate": 5.089627554190307e-06, + "loss": 0.0121, + "step": 8920 + }, + { + "epoch": 14.958123953098827, + "grad_norm": 0.0604483038187027, + "learning_rate": 5.058635485049113e-06, + "loss": 0.0142, + "step": 8930 + }, + { + "epoch": 14.974874371859297, + "grad_norm": 0.08353114128112793, + "learning_rate": 5.027718913963378e-06, + "loss": 0.0098, + "step": 8940 + }, + { + "epoch": 14.991624790619765, + "grad_norm": 0.08845911175012589, + "learning_rate": 4.996878075723912e-06, + "loss": 0.0112, + "step": 8950 + }, + { + "epoch": 15.008375209380235, + "grad_norm": 0.059438783675432205, + "learning_rate": 4.966113204546377e-06, + "loss": 0.011, + "step": 8960 + }, + { + "epoch": 15.025125628140703, + "grad_norm": 0.0784023180603981, + "learning_rate": 4.93542453406952e-06, + "loss": 0.013, + "step": 8970 + }, + { + "epoch": 15.041876046901173, + "grad_norm": 0.08503511548042297, + "learning_rate": 4.904812297353403e-06, + "loss": 0.0118, + "step": 8980 + }, + { + "epoch": 15.058626465661641, + "grad_norm": 0.08692113310098648, + "learning_rate": 4.874276726877604e-06, + "loss": 0.0142, + "step": 8990 + }, + { + "epoch": 15.075376884422111, + "grad_norm": 0.08364814519882202, + "learning_rate": 4.843818054539489e-06, + "loss": 0.0112, + "step": 9000 + }, + { + "epoch": 15.09212730318258, + "grad_norm": 0.07530484348535538, + "learning_rate": 4.813436511652433e-06, + "loss": 0.0123, + "step": 9010 + }, + { + "epoch": 15.108877721943049, + "grad_norm": 0.0713859349489212, + "learning_rate": 4.7831323289440525e-06, + "loss": 0.0115, + "step": 9020 + }, + { + "epoch": 15.125628140703517, + "grad_norm": 0.06153390556573868, + "learning_rate": 4.752905736554474e-06, + "loss": 0.0118, + "step": 9030 + }, + { + "epoch": 15.142378559463987, + "grad_norm": 0.06609758734703064, + "learning_rate": 4.72275696403458e-06, + "loss": 0.0105, + "step": 9040 + }, + { + "epoch": 15.159128978224455, + "grad_norm": 0.05786735191941261, + "learning_rate": 4.6926862403442495e-06, + "loss": 0.0103, + "step": 9050 + }, + { + "epoch": 15.175879396984925, + "grad_norm": 0.06739120185375214, + "learning_rate": 4.662693793850643e-06, + "loss": 0.0111, + "step": 9060 + }, + { + "epoch": 15.192629815745393, + "grad_norm": 0.06687992811203003, + "learning_rate": 4.632779852326465e-06, + "loss": 0.011, + "step": 9070 + }, + { + "epoch": 15.209380234505863, + "grad_norm": 0.11040312796831131, + "learning_rate": 4.6029446429482055e-06, + "loss": 0.0124, + "step": 9080 + }, + { + "epoch": 15.226130653266331, + "grad_norm": 0.08872684091329575, + "learning_rate": 4.573188392294452e-06, + "loss": 0.0108, + "step": 9090 + }, + { + "epoch": 15.242881072026801, + "grad_norm": 0.07327628880739212, + "learning_rate": 4.543511326344155e-06, + "loss": 0.013, + "step": 9100 + }, + { + "epoch": 15.25963149078727, + "grad_norm": 0.057703230530023575, + "learning_rate": 4.513913670474896e-06, + "loss": 0.009, + "step": 9110 + }, + { + "epoch": 15.27638190954774, + "grad_norm": 0.0857810378074646, + "learning_rate": 4.4843956494612e-06, + "loss": 0.0115, + "step": 9120 + }, + { + "epoch": 15.293132328308207, + "grad_norm": 0.05844645947217941, + "learning_rate": 4.454957487472821e-06, + "loss": 0.013, + "step": 9130 + }, + { + "epoch": 15.309882747068677, + "grad_norm": 0.07501334697008133, + "learning_rate": 4.4255994080730236e-06, + "loss": 0.0102, + "step": 9140 + }, + { + "epoch": 15.326633165829145, + "grad_norm": 0.09073469042778015, + "learning_rate": 4.396321634216911e-06, + "loss": 0.0116, + "step": 9150 + }, + { + "epoch": 15.343383584589615, + "grad_norm": 0.06694762408733368, + "learning_rate": 4.367124388249717e-06, + "loss": 0.0112, + "step": 9160 + }, + { + "epoch": 15.360134003350083, + "grad_norm": 0.06576213240623474, + "learning_rate": 4.338007891905113e-06, + "loss": 0.0102, + "step": 9170 + }, + { + "epoch": 15.376884422110553, + "grad_norm": 0.05746903270483017, + "learning_rate": 4.308972366303536e-06, + "loss": 0.0093, + "step": 9180 + }, + { + "epoch": 15.393634840871021, + "grad_norm": 0.06974180042743683, + "learning_rate": 4.280018031950506e-06, + "loss": 0.0124, + "step": 9190 + }, + { + "epoch": 15.410385259631491, + "grad_norm": 0.08683093637228012, + "learning_rate": 4.251145108734941e-06, + "loss": 0.0125, + "step": 9200 + }, + { + "epoch": 15.42713567839196, + "grad_norm": 0.05769151821732521, + "learning_rate": 4.222353815927505e-06, + "loss": 0.0113, + "step": 9210 + }, + { + "epoch": 15.44388609715243, + "grad_norm": 0.06937823444604874, + "learning_rate": 4.193644372178931e-06, + "loss": 0.0114, + "step": 9220 + }, + { + "epoch": 15.460636515912897, + "grad_norm": 0.06178734824061394, + "learning_rate": 4.1650169955183535e-06, + "loss": 0.011, + "step": 9230 + }, + { + "epoch": 15.477386934673367, + "grad_norm": 0.055109817534685135, + "learning_rate": 4.136471903351681e-06, + "loss": 0.0099, + "step": 9240 + }, + { + "epoch": 15.494137353433835, + "grad_norm": 0.08306293189525604, + "learning_rate": 4.108009312459902e-06, + "loss": 0.011, + "step": 9250 + }, + { + "epoch": 15.510887772194305, + "grad_norm": 0.07963228225708008, + "learning_rate": 4.079629438997487e-06, + "loss": 0.0128, + "step": 9260 + }, + { + "epoch": 15.527638190954773, + "grad_norm": 0.08196922391653061, + "learning_rate": 4.051332498490699e-06, + "loss": 0.0118, + "step": 9270 + }, + { + "epoch": 15.544388609715243, + "grad_norm": 0.07916904240846634, + "learning_rate": 4.0231187058359996e-06, + "loss": 0.0126, + "step": 9280 + }, + { + "epoch": 15.561139028475711, + "grad_norm": 0.05994735658168793, + "learning_rate": 3.994988275298393e-06, + "loss": 0.0113, + "step": 9290 + }, + { + "epoch": 15.577889447236181, + "grad_norm": 0.07669159024953842, + "learning_rate": 3.966941420509793e-06, + "loss": 0.0118, + "step": 9300 + }, + { + "epoch": 15.59463986599665, + "grad_norm": 0.08148621767759323, + "learning_rate": 3.9389783544674215e-06, + "loss": 0.0113, + "step": 9310 + }, + { + "epoch": 15.61139028475712, + "grad_norm": 0.06318013370037079, + "learning_rate": 3.911099289532183e-06, + "loss": 0.0098, + "step": 9320 + }, + { + "epoch": 15.628140703517587, + "grad_norm": 0.08033207803964615, + "learning_rate": 3.883304437427038e-06, + "loss": 0.0103, + "step": 9330 + }, + { + "epoch": 15.644891122278057, + "grad_norm": 0.06583742052316666, + "learning_rate": 3.855594009235415e-06, + "loss": 0.0133, + "step": 9340 + }, + { + "epoch": 15.661641541038525, + "grad_norm": 0.0734059065580368, + "learning_rate": 3.827968215399601e-06, + "loss": 0.011, + "step": 9350 + }, + { + "epoch": 15.678391959798995, + "grad_norm": 0.059179969131946564, + "learning_rate": 3.800427265719131e-06, + "loss": 0.0124, + "step": 9360 + }, + { + "epoch": 15.695142378559463, + "grad_norm": 0.07946020364761353, + "learning_rate": 3.7729713693492142e-06, + "loss": 0.0103, + "step": 9370 + }, + { + "epoch": 15.711892797319933, + "grad_norm": 0.07202473282814026, + "learning_rate": 3.7456007347991363e-06, + "loss": 0.0107, + "step": 9380 + }, + { + "epoch": 15.728643216080402, + "grad_norm": 0.06295417994260788, + "learning_rate": 3.7183155699306686e-06, + "loss": 0.0104, + "step": 9390 + }, + { + "epoch": 15.745393634840871, + "grad_norm": 0.10039746016263962, + "learning_rate": 3.691116081956502e-06, + "loss": 0.0124, + "step": 9400 + }, + { + "epoch": 15.76214405360134, + "grad_norm": 0.06262555718421936, + "learning_rate": 3.6640024774386754e-06, + "loss": 0.0111, + "step": 9410 + }, + { + "epoch": 15.77889447236181, + "grad_norm": 0.06782974302768707, + "learning_rate": 3.636974962286983e-06, + "loss": 0.0096, + "step": 9420 + }, + { + "epoch": 15.795644891122278, + "grad_norm": 0.07636243849992752, + "learning_rate": 3.610033741757442e-06, + "loss": 0.0106, + "step": 9430 + }, + { + "epoch": 15.812395309882747, + "grad_norm": 0.06166220083832741, + "learning_rate": 3.5831790204507185e-06, + "loss": 0.009, + "step": 9440 + }, + { + "epoch": 15.829145728643216, + "grad_norm": 0.0528729185461998, + "learning_rate": 3.5564110023105602e-06, + "loss": 0.0106, + "step": 9450 + }, + { + "epoch": 15.845896147403685, + "grad_norm": 0.07870613783597946, + "learning_rate": 3.5297298906222796e-06, + "loss": 0.0117, + "step": 9460 + }, + { + "epoch": 15.862646566164154, + "grad_norm": 0.07957785576581955, + "learning_rate": 3.5031358880111873e-06, + "loss": 0.0124, + "step": 9470 + }, + { + "epoch": 15.879396984924623, + "grad_norm": 0.07852274179458618, + "learning_rate": 3.4766291964410474e-06, + "loss": 0.0097, + "step": 9480 + }, + { + "epoch": 15.896147403685092, + "grad_norm": 0.06998380273580551, + "learning_rate": 3.450210017212569e-06, + "loss": 0.012, + "step": 9490 + }, + { + "epoch": 15.912897822445562, + "grad_norm": 0.0631658136844635, + "learning_rate": 3.4238785509618596e-06, + "loss": 0.0087, + "step": 9500 + }, + { + "epoch": 15.92964824120603, + "grad_norm": 0.09431459754705429, + "learning_rate": 3.3976349976588956e-06, + "loss": 0.0107, + "step": 9510 + }, + { + "epoch": 15.9463986599665, + "grad_norm": 0.06583735346794128, + "learning_rate": 3.3714795566060256e-06, + "loss": 0.0119, + "step": 9520 + }, + { + "epoch": 15.963149078726968, + "grad_norm": 0.09137707203626633, + "learning_rate": 3.3454124264364424e-06, + "loss": 0.0095, + "step": 9530 + }, + { + "epoch": 15.979899497487438, + "grad_norm": 0.07819409668445587, + "learning_rate": 3.3194338051126626e-06, + "loss": 0.011, + "step": 9540 + }, + { + "epoch": 15.996649916247906, + "grad_norm": 0.059675078839063644, + "learning_rate": 3.293543889925053e-06, + "loss": 0.0103, + "step": 9550 + }, + { + "epoch": 16.013400335008374, + "grad_norm": 0.08947017788887024, + "learning_rate": 3.2677428774903144e-06, + "loss": 0.0099, + "step": 9560 + }, + { + "epoch": 16.030150753768844, + "grad_norm": 0.06796885281801224, + "learning_rate": 3.2420309637499768e-06, + "loss": 0.0121, + "step": 9570 + }, + { + "epoch": 16.046901172529314, + "grad_norm": 0.10741371661424637, + "learning_rate": 3.2164083439689406e-06, + "loss": 0.0111, + "step": 9580 + }, + { + "epoch": 16.063651591289783, + "grad_norm": 0.08767179399728775, + "learning_rate": 3.1908752127339745e-06, + "loss": 0.0114, + "step": 9590 + }, + { + "epoch": 16.08040201005025, + "grad_norm": 0.0646979808807373, + "learning_rate": 3.1654317639522306e-06, + "loss": 0.0106, + "step": 9600 + }, + { + "epoch": 16.09715242881072, + "grad_norm": 0.07663914561271667, + "learning_rate": 3.140078190849794e-06, + "loss": 0.0107, + "step": 9610 + }, + { + "epoch": 16.11390284757119, + "grad_norm": 0.05584222078323364, + "learning_rate": 3.114814685970204e-06, + "loss": 0.0105, + "step": 9620 + }, + { + "epoch": 16.13065326633166, + "grad_norm": 0.07266643643379211, + "learning_rate": 3.0896414411729784e-06, + "loss": 0.012, + "step": 9630 + }, + { + "epoch": 16.147403685092126, + "grad_norm": 0.12060057371854782, + "learning_rate": 3.064558647632188e-06, + "loss": 0.0114, + "step": 9640 + }, + { + "epoch": 16.164154103852596, + "grad_norm": 0.07776009291410446, + "learning_rate": 3.0395664958349684e-06, + "loss": 0.0116, + "step": 9650 + }, + { + "epoch": 16.180904522613066, + "grad_norm": 0.06058536469936371, + "learning_rate": 3.0146651755801097e-06, + "loss": 0.009, + "step": 9660 + }, + { + "epoch": 16.197654941373536, + "grad_norm": 0.08329997956752777, + "learning_rate": 2.989854875976577e-06, + "loss": 0.01, + "step": 9670 + }, + { + "epoch": 16.214405360134002, + "grad_norm": 0.05222083628177643, + "learning_rate": 2.96513578544211e-06, + "loss": 0.0112, + "step": 9680 + }, + { + "epoch": 16.231155778894472, + "grad_norm": 0.07646694034337997, + "learning_rate": 2.9405080917017747e-06, + "loss": 0.0107, + "step": 9690 + }, + { + "epoch": 16.24790619765494, + "grad_norm": 0.08242721110582352, + "learning_rate": 2.915971981786526e-06, + "loss": 0.0113, + "step": 9700 + }, + { + "epoch": 16.26465661641541, + "grad_norm": 0.06174449622631073, + "learning_rate": 2.891527642031815e-06, + "loss": 0.0115, + "step": 9710 + }, + { + "epoch": 16.281407035175878, + "grad_norm": 0.07003425806760788, + "learning_rate": 2.8671752580761555e-06, + "loss": 0.0098, + "step": 9720 + }, + { + "epoch": 16.298157453936348, + "grad_norm": 0.045007798820734024, + "learning_rate": 2.8429150148597117e-06, + "loss": 0.0101, + "step": 9730 + }, + { + "epoch": 16.314907872696818, + "grad_norm": 0.06540249288082123, + "learning_rate": 2.8187470966229054e-06, + "loss": 0.0111, + "step": 9740 + }, + { + "epoch": 16.331658291457288, + "grad_norm": 0.05390077829360962, + "learning_rate": 2.7946716869050197e-06, + "loss": 0.0141, + "step": 9750 + }, + { + "epoch": 16.348408710217754, + "grad_norm": 0.09093419462442398, + "learning_rate": 2.770688968542777e-06, + "loss": 0.0106, + "step": 9760 + }, + { + "epoch": 16.365159128978224, + "grad_norm": 0.07020417600870132, + "learning_rate": 2.746799123668985e-06, + "loss": 0.0095, + "step": 9770 + }, + { + "epoch": 16.381909547738694, + "grad_norm": 0.0639248639345169, + "learning_rate": 2.7230023337111425e-06, + "loss": 0.0106, + "step": 9780 + }, + { + "epoch": 16.398659966499164, + "grad_norm": 0.07816614955663681, + "learning_rate": 2.6992987793900448e-06, + "loss": 0.0121, + "step": 9790 + }, + { + "epoch": 16.41541038525963, + "grad_norm": 0.059835903346538544, + "learning_rate": 2.6756886407184313e-06, + "loss": 0.0095, + "step": 9800 + }, + { + "epoch": 16.4321608040201, + "grad_norm": 0.07151816040277481, + "learning_rate": 2.6521720969996172e-06, + "loss": 0.0101, + "step": 9810 + }, + { + "epoch": 16.44891122278057, + "grad_norm": 0.09019379317760468, + "learning_rate": 2.6287493268261137e-06, + "loss": 0.0113, + "step": 9820 + }, + { + "epoch": 16.46566164154104, + "grad_norm": 0.08206336200237274, + "learning_rate": 2.6054205080782946e-06, + "loss": 0.0098, + "step": 9830 + }, + { + "epoch": 16.482412060301506, + "grad_norm": 0.06948984414339066, + "learning_rate": 2.582185817923034e-06, + "loss": 0.0112, + "step": 9840 + }, + { + "epoch": 16.499162479061976, + "grad_norm": 0.07291114330291748, + "learning_rate": 2.5590454328123538e-06, + "loss": 0.0102, + "step": 9850 + }, + { + "epoch": 16.515912897822446, + "grad_norm": 0.0553848072886467, + "learning_rate": 2.5359995284820993e-06, + "loss": 0.0118, + "step": 9860 + }, + { + "epoch": 16.532663316582916, + "grad_norm": 0.08315521478652954, + "learning_rate": 2.513048279950602e-06, + "loss": 0.0111, + "step": 9870 + }, + { + "epoch": 16.549413735343382, + "grad_norm": 0.06910669803619385, + "learning_rate": 2.4901918615173274e-06, + "loss": 0.0104, + "step": 9880 + }, + { + "epoch": 16.566164154103852, + "grad_norm": 0.09397564828395844, + "learning_rate": 2.4674304467615876e-06, + "loss": 0.0116, + "step": 9890 + }, + { + "epoch": 16.582914572864322, + "grad_norm": 0.05459596961736679, + "learning_rate": 2.4447642085412007e-06, + "loss": 0.0115, + "step": 9900 + }, + { + "epoch": 16.599664991624792, + "grad_norm": 0.06284655630588531, + "learning_rate": 2.4221933189911734e-06, + "loss": 0.0102, + "step": 9910 + }, + { + "epoch": 16.616415410385258, + "grad_norm": 0.09468510001897812, + "learning_rate": 2.3997179495224104e-06, + "loss": 0.0123, + "step": 9920 + }, + { + "epoch": 16.633165829145728, + "grad_norm": 0.06382662802934647, + "learning_rate": 2.3773382708204096e-06, + "loss": 0.0146, + "step": 9930 + }, + { + "epoch": 16.649916247906198, + "grad_norm": 0.07933798432350159, + "learning_rate": 2.3550544528439443e-06, + "loss": 0.0107, + "step": 9940 + }, + { + "epoch": 16.666666666666668, + "grad_norm": 0.06377198547124863, + "learning_rate": 2.3328666648238076e-06, + "loss": 0.0104, + "step": 9950 + }, + { + "epoch": 16.683417085427134, + "grad_norm": 0.06982989609241486, + "learning_rate": 2.3107750752615027e-06, + "loss": 0.0093, + "step": 9960 + }, + { + "epoch": 16.700167504187604, + "grad_norm": 0.054972875863313675, + "learning_rate": 2.288779851927961e-06, + "loss": 0.0095, + "step": 9970 + }, + { + "epoch": 16.716917922948074, + "grad_norm": 0.06242859736084938, + "learning_rate": 2.266881161862288e-06, + "loss": 0.0107, + "step": 9980 + }, + { + "epoch": 16.733668341708544, + "grad_norm": 0.07019710540771484, + "learning_rate": 2.2450791713704806e-06, + "loss": 0.0102, + "step": 9990 + }, + { + "epoch": 16.75041876046901, + "grad_norm": 0.047752682119607925, + "learning_rate": 2.2233740460241593e-06, + "loss": 0.0111, + "step": 10000 + }, + { + "epoch": 16.76716917922948, + "grad_norm": 0.090115487575531, + "learning_rate": 2.2017659506593256e-06, + "loss": 0.0099, + "step": 10010 + }, + { + "epoch": 16.78391959798995, + "grad_norm": 0.06850699335336685, + "learning_rate": 2.1802550493751043e-06, + "loss": 0.0105, + "step": 10020 + }, + { + "epoch": 16.80067001675042, + "grad_norm": 0.08048104494810104, + "learning_rate": 2.158841505532488e-06, + "loss": 0.0104, + "step": 10030 + }, + { + "epoch": 16.817420435510886, + "grad_norm": 0.06437097489833832, + "learning_rate": 2.1375254817531104e-06, + "loss": 0.0118, + "step": 10040 + }, + { + "epoch": 16.834170854271356, + "grad_norm": 0.07399899512529373, + "learning_rate": 2.1163071399180106e-06, + "loss": 0.0096, + "step": 10050 + }, + { + "epoch": 16.850921273031826, + "grad_norm": 0.08019571006298065, + "learning_rate": 2.0951866411663746e-06, + "loss": 0.0101, + "step": 10060 + }, + { + "epoch": 16.867671691792296, + "grad_norm": 0.06191406771540642, + "learning_rate": 2.0741641458943633e-06, + "loss": 0.0095, + "step": 10070 + }, + { + "epoch": 16.884422110552762, + "grad_norm": 0.06613759696483612, + "learning_rate": 2.053239813753854e-06, + "loss": 0.0112, + "step": 10080 + }, + { + "epoch": 16.901172529313232, + "grad_norm": 0.05650663748383522, + "learning_rate": 2.0324138036512358e-06, + "loss": 0.011, + "step": 10090 + }, + { + "epoch": 16.917922948073702, + "grad_norm": 0.08308195322751999, + "learning_rate": 2.011686273746216e-06, + "loss": 0.011, + "step": 10100 + }, + { + "epoch": 16.934673366834172, + "grad_norm": 0.08628930151462555, + "learning_rate": 1.991057381450608e-06, + "loss": 0.0128, + "step": 10110 + }, + { + "epoch": 16.95142378559464, + "grad_norm": 0.07573705166578293, + "learning_rate": 1.9705272834271337e-06, + "loss": 0.0114, + "step": 10120 + }, + { + "epoch": 16.96817420435511, + "grad_norm": 0.056926365941762924, + "learning_rate": 1.9500961355882418e-06, + "loss": 0.0097, + "step": 10130 + }, + { + "epoch": 16.984924623115578, + "grad_norm": 0.06239789351820946, + "learning_rate": 1.9297640930949217e-06, + "loss": 0.0098, + "step": 10140 + }, + { + "epoch": 17.001675041876048, + "grad_norm": 0.06450553983449936, + "learning_rate": 1.909531310355518e-06, + "loss": 0.0085, + "step": 10150 + }, + { + "epoch": 17.018425460636514, + "grad_norm": 0.07283646613359451, + "learning_rate": 1.8893979410245694e-06, + "loss": 0.0104, + "step": 10160 + }, + { + "epoch": 17.035175879396984, + "grad_norm": 0.07477754354476929, + "learning_rate": 1.8693641380016341e-06, + "loss": 0.0106, + "step": 10170 + }, + { + "epoch": 17.051926298157454, + "grad_norm": 0.0838644728064537, + "learning_rate": 1.8494300534301268e-06, + "loss": 0.0119, + "step": 10180 + }, + { + "epoch": 17.068676716917924, + "grad_norm": 0.07156196236610413, + "learning_rate": 1.8295958386961691e-06, + "loss": 0.0096, + "step": 10190 + }, + { + "epoch": 17.08542713567839, + "grad_norm": 0.052184104919433594, + "learning_rate": 1.8098616444274435e-06, + "loss": 0.0104, + "step": 10200 + }, + { + "epoch": 17.10217755443886, + "grad_norm": 0.07056842744350433, + "learning_rate": 1.7902276204920326e-06, + "loss": 0.0108, + "step": 10210 + }, + { + "epoch": 17.11892797319933, + "grad_norm": 0.06949982792139053, + "learning_rate": 1.770693915997298e-06, + "loss": 0.0113, + "step": 10220 + }, + { + "epoch": 17.1356783919598, + "grad_norm": 0.06455013900995255, + "learning_rate": 1.7512606792887436e-06, + "loss": 0.0095, + "step": 10230 + }, + { + "epoch": 17.152428810720266, + "grad_norm": 0.06968142837285995, + "learning_rate": 1.7319280579488877e-06, + "loss": 0.0103, + "step": 10240 + }, + { + "epoch": 17.169179229480736, + "grad_norm": 0.07281049340963364, + "learning_rate": 1.712696198796131e-06, + "loss": 0.0098, + "step": 10250 + }, + { + "epoch": 17.185929648241206, + "grad_norm": 0.0891086608171463, + "learning_rate": 1.6935652478836644e-06, + "loss": 0.011, + "step": 10260 + }, + { + "epoch": 17.202680067001676, + "grad_norm": 0.06211129575967789, + "learning_rate": 1.6745353504983473e-06, + "loss": 0.0108, + "step": 10270 + }, + { + "epoch": 17.219430485762143, + "grad_norm": 0.08370804041624069, + "learning_rate": 1.6556066511595947e-06, + "loss": 0.0105, + "step": 10280 + }, + { + "epoch": 17.236180904522612, + "grad_norm": 0.0769832655787468, + "learning_rate": 1.636779293618299e-06, + "loss": 0.0126, + "step": 10290 + }, + { + "epoch": 17.252931323283082, + "grad_norm": 0.06175370141863823, + "learning_rate": 1.618053420855729e-06, + "loss": 0.0094, + "step": 10300 + }, + { + "epoch": 17.269681742043552, + "grad_norm": 0.08861753344535828, + "learning_rate": 1.5994291750824364e-06, + "loss": 0.0117, + "step": 10310 + }, + { + "epoch": 17.28643216080402, + "grad_norm": 0.057883117347955704, + "learning_rate": 1.58090669773719e-06, + "loss": 0.0096, + "step": 10320 + }, + { + "epoch": 17.30318257956449, + "grad_norm": 0.05262555181980133, + "learning_rate": 1.5624861294858971e-06, + "loss": 0.0098, + "step": 10330 + }, + { + "epoch": 17.31993299832496, + "grad_norm": 0.07395872473716736, + "learning_rate": 1.5441676102205244e-06, + "loss": 0.0103, + "step": 10340 + }, + { + "epoch": 17.33668341708543, + "grad_norm": 0.06096988916397095, + "learning_rate": 1.5259512790580493e-06, + "loss": 0.0105, + "step": 10350 + }, + { + "epoch": 17.353433835845895, + "grad_norm": 0.05924631655216217, + "learning_rate": 1.5078372743394043e-06, + "loss": 0.0104, + "step": 10360 + }, + { + "epoch": 17.370184254606365, + "grad_norm": 0.0643719881772995, + "learning_rate": 1.4898257336284078e-06, + "loss": 0.0114, + "step": 10370 + }, + { + "epoch": 17.386934673366834, + "grad_norm": 0.07372860610485077, + "learning_rate": 1.4719167937107425e-06, + "loss": 0.0093, + "step": 10380 + }, + { + "epoch": 17.403685092127304, + "grad_norm": 0.0642753317952156, + "learning_rate": 1.4541105905929047e-06, + "loss": 0.0102, + "step": 10390 + }, + { + "epoch": 17.42043551088777, + "grad_norm": 0.09747152030467987, + "learning_rate": 1.4364072595011679e-06, + "loss": 0.0134, + "step": 10400 + }, + { + "epoch": 17.43718592964824, + "grad_norm": 0.0734635517001152, + "learning_rate": 1.4188069348805655e-06, + "loss": 0.0102, + "step": 10410 + }, + { + "epoch": 17.45393634840871, + "grad_norm": 0.05024496838450432, + "learning_rate": 1.4013097503938682e-06, + "loss": 0.0107, + "step": 10420 + }, + { + "epoch": 17.47068676716918, + "grad_norm": 0.0604495108127594, + "learning_rate": 1.3839158389205536e-06, + "loss": 0.0096, + "step": 10430 + }, + { + "epoch": 17.487437185929647, + "grad_norm": 0.06331069022417068, + "learning_rate": 1.3666253325558232e-06, + "loss": 0.0084, + "step": 10440 + }, + { + "epoch": 17.504187604690117, + "grad_norm": 0.08753936737775803, + "learning_rate": 1.3494383626095797e-06, + "loss": 0.01, + "step": 10450 + }, + { + "epoch": 17.520938023450586, + "grad_norm": 0.0691802129149437, + "learning_rate": 1.3323550596054368e-06, + "loss": 0.0102, + "step": 10460 + }, + { + "epoch": 17.537688442211056, + "grad_norm": 0.06930302828550339, + "learning_rate": 1.3153755532797196e-06, + "loss": 0.0105, + "step": 10470 + }, + { + "epoch": 17.554438860971523, + "grad_norm": 0.11355336755514145, + "learning_rate": 1.2984999725804986e-06, + "loss": 0.01, + "step": 10480 + }, + { + "epoch": 17.571189279731993, + "grad_norm": 0.08153529465198517, + "learning_rate": 1.2817284456665856e-06, + "loss": 0.0121, + "step": 10490 + }, + { + "epoch": 17.587939698492463, + "grad_norm": 0.05092867091298103, + "learning_rate": 1.2650610999065849e-06, + "loss": 0.011, + "step": 10500 + }, + { + "epoch": 17.604690117252932, + "grad_norm": 0.06719738990068436, + "learning_rate": 1.2484980618779097e-06, + "loss": 0.0111, + "step": 10510 + }, + { + "epoch": 17.6214405360134, + "grad_norm": 0.07338828593492508, + "learning_rate": 1.2320394573658188e-06, + "loss": 0.0096, + "step": 10520 + }, + { + "epoch": 17.63819095477387, + "grad_norm": 0.10995873808860779, + "learning_rate": 1.215685411362476e-06, + "loss": 0.011, + "step": 10530 + }, + { + "epoch": 17.65494137353434, + "grad_norm": 0.060237374156713486, + "learning_rate": 1.1994360480659945e-06, + "loss": 0.0107, + "step": 10540 + }, + { + "epoch": 17.67169179229481, + "grad_norm": 0.07575573772192001, + "learning_rate": 1.18329149087948e-06, + "loss": 0.0121, + "step": 10550 + }, + { + "epoch": 17.688442211055275, + "grad_norm": 0.05417560040950775, + "learning_rate": 1.1672518624101176e-06, + "loss": 0.0099, + "step": 10560 + }, + { + "epoch": 17.705192629815745, + "grad_norm": 0.05768902972340584, + "learning_rate": 1.1513172844682251e-06, + "loss": 0.0096, + "step": 10570 + }, + { + "epoch": 17.721943048576215, + "grad_norm": 0.07235273718833923, + "learning_rate": 1.1354878780663263e-06, + "loss": 0.0115, + "step": 10580 + }, + { + "epoch": 17.738693467336685, + "grad_norm": 0.06070183590054512, + "learning_rate": 1.1197637634182433e-06, + "loss": 0.0111, + "step": 10590 + }, + { + "epoch": 17.75544388609715, + "grad_norm": 0.06319398432970047, + "learning_rate": 1.1041450599381765e-06, + "loss": 0.0087, + "step": 10600 + }, + { + "epoch": 17.77219430485762, + "grad_norm": 0.06955882161855698, + "learning_rate": 1.0886318862397926e-06, + "loss": 0.0098, + "step": 10610 + }, + { + "epoch": 17.78894472361809, + "grad_norm": 0.07007668912410736, + "learning_rate": 1.073224360135337e-06, + "loss": 0.0124, + "step": 10620 + }, + { + "epoch": 17.80569514237856, + "grad_norm": 0.06133275106549263, + "learning_rate": 1.0579225986347312e-06, + "loss": 0.0105, + "step": 10630 + }, + { + "epoch": 17.822445561139027, + "grad_norm": 0.05876230075955391, + "learning_rate": 1.042726717944678e-06, + "loss": 0.0128, + "step": 10640 + }, + { + "epoch": 17.839195979899497, + "grad_norm": 0.06510904431343079, + "learning_rate": 1.0276368334677932e-06, + "loss": 0.0105, + "step": 10650 + }, + { + "epoch": 17.855946398659967, + "grad_norm": 0.08215858787298203, + "learning_rate": 1.0126530598017215e-06, + "loss": 0.0099, + "step": 10660 + }, + { + "epoch": 17.872696817420437, + "grad_norm": 0.07347946614027023, + "learning_rate": 9.977755107382586e-07, + "loss": 0.0094, + "step": 10670 + }, + { + "epoch": 17.889447236180903, + "grad_norm": 0.05123291164636612, + "learning_rate": 9.830042992625027e-07, + "loss": 0.0106, + "step": 10680 + }, + { + "epoch": 17.906197654941373, + "grad_norm": 0.06298977136611938, + "learning_rate": 9.683395375519911e-07, + "loss": 0.0116, + "step": 10690 + }, + { + "epoch": 17.922948073701843, + "grad_norm": 0.08945272117853165, + "learning_rate": 9.537813369758359e-07, + "loss": 0.0108, + "step": 10700 + }, + { + "epoch": 17.939698492462313, + "grad_norm": 0.08734112977981567, + "learning_rate": 9.393298080938934e-07, + "loss": 0.0105, + "step": 10710 + }, + { + "epoch": 17.95644891122278, + "grad_norm": 0.08413497358560562, + "learning_rate": 9.249850606559246e-07, + "loss": 0.0105, + "step": 10720 + }, + { + "epoch": 17.97319932998325, + "grad_norm": 0.06129211187362671, + "learning_rate": 9.107472036007458e-07, + "loss": 0.0117, + "step": 10730 + }, + { + "epoch": 17.98994974874372, + "grad_norm": 0.05145135521888733, + "learning_rate": 8.96616345055421e-07, + "loss": 0.01, + "step": 10740 + }, + { + "epoch": 18.00670016750419, + "grad_norm": 0.04860294237732887, + "learning_rate": 8.825925923344275e-07, + "loss": 0.011, + "step": 10750 + }, + { + "epoch": 18.023450586264655, + "grad_norm": 0.06857185065746307, + "learning_rate": 8.686760519388432e-07, + "loss": 0.0085, + "step": 10760 + }, + { + "epoch": 18.040201005025125, + "grad_norm": 0.051961518824100494, + "learning_rate": 8.548668295555439e-07, + "loss": 0.0096, + "step": 10770 + }, + { + "epoch": 18.056951423785595, + "grad_norm": 0.0770072340965271, + "learning_rate": 8.411650300563961e-07, + "loss": 0.0091, + "step": 10780 + }, + { + "epoch": 18.073701842546065, + "grad_norm": 0.06823989003896713, + "learning_rate": 8.275707574974551e-07, + "loss": 0.0091, + "step": 10790 + }, + { + "epoch": 18.09045226130653, + "grad_norm": 0.08679325133562088, + "learning_rate": 8.140841151181894e-07, + "loss": 0.012, + "step": 10800 + }, + { + "epoch": 18.107202680067, + "grad_norm": 0.0784691721200943, + "learning_rate": 8.007052053406816e-07, + "loss": 0.0116, + "step": 10810 + }, + { + "epoch": 18.12395309882747, + "grad_norm": 0.07049158215522766, + "learning_rate": 7.87434129768857e-07, + "loss": 0.0096, + "step": 10820 + }, + { + "epoch": 18.14070351758794, + "grad_norm": 0.06313124299049377, + "learning_rate": 7.742709891877159e-07, + "loss": 0.0103, + "step": 10830 + }, + { + "epoch": 18.157453936348407, + "grad_norm": 0.09241461008787155, + "learning_rate": 7.61215883562561e-07, + "loss": 0.0099, + "step": 10840 + }, + { + "epoch": 18.174204355108877, + "grad_norm": 0.06502427905797958, + "learning_rate": 7.482689120382397e-07, + "loss": 0.0112, + "step": 10850 + }, + { + "epoch": 18.190954773869347, + "grad_norm": 0.076650470495224, + "learning_rate": 7.354301729383944e-07, + "loss": 0.011, + "step": 10860 + }, + { + "epoch": 18.207705192629817, + "grad_norm": 0.07330027967691422, + "learning_rate": 7.226997637647204e-07, + "loss": 0.0107, + "step": 10870 + }, + { + "epoch": 18.224455611390283, + "grad_norm": 0.05841333046555519, + "learning_rate": 7.100777811962072e-07, + "loss": 0.0107, + "step": 10880 + }, + { + "epoch": 18.241206030150753, + "grad_norm": 0.07078751176595688, + "learning_rate": 6.975643210884252e-07, + "loss": 0.0098, + "step": 10890 + }, + { + "epoch": 18.257956448911223, + "grad_norm": 0.05953970178961754, + "learning_rate": 6.851594784727838e-07, + "loss": 0.0106, + "step": 10900 + }, + { + "epoch": 18.274706867671693, + "grad_norm": 0.0576588436961174, + "learning_rate": 6.728633475558221e-07, + "loss": 0.0112, + "step": 10910 + }, + { + "epoch": 18.29145728643216, + "grad_norm": 0.058967769145965576, + "learning_rate": 6.606760217184766e-07, + "loss": 0.0115, + "step": 10920 + }, + { + "epoch": 18.30820770519263, + "grad_norm": 0.057810813188552856, + "learning_rate": 6.485975935153877e-07, + "loss": 0.0109, + "step": 10930 + }, + { + "epoch": 18.3249581239531, + "grad_norm": 0.07053758949041367, + "learning_rate": 6.366281546741909e-07, + "loss": 0.0103, + "step": 10940 + }, + { + "epoch": 18.34170854271357, + "grad_norm": 0.06809135526418686, + "learning_rate": 6.247677960948139e-07, + "loss": 0.0103, + "step": 10950 + }, + { + "epoch": 18.358458961474035, + "grad_norm": 0.05694377422332764, + "learning_rate": 6.130166078487987e-07, + "loss": 0.0111, + "step": 10960 + }, + { + "epoch": 18.375209380234505, + "grad_norm": 0.09857413917779922, + "learning_rate": 6.013746791786117e-07, + "loss": 0.011, + "step": 10970 + }, + { + "epoch": 18.391959798994975, + "grad_norm": 0.07620888203382492, + "learning_rate": 5.898420984969588e-07, + "loss": 0.0103, + "step": 10980 + }, + { + "epoch": 18.408710217755445, + "grad_norm": 0.06404457986354828, + "learning_rate": 5.784189533861262e-07, + "loss": 0.0086, + "step": 10990 + }, + { + "epoch": 18.42546063651591, + "grad_norm": 0.06898780912160873, + "learning_rate": 5.671053305973084e-07, + "loss": 0.0134, + "step": 11000 + }, + { + "epoch": 18.44221105527638, + "grad_norm": 0.0845097228884697, + "learning_rate": 5.559013160499504e-07, + "loss": 0.0107, + "step": 11010 + }, + { + "epoch": 18.45896147403685, + "grad_norm": 0.07644879817962646, + "learning_rate": 5.44806994831093e-07, + "loss": 0.0111, + "step": 11020 + }, + { + "epoch": 18.47571189279732, + "grad_norm": 0.0650167316198349, + "learning_rate": 5.338224511947348e-07, + "loss": 0.0103, + "step": 11030 + }, + { + "epoch": 18.492462311557787, + "grad_norm": 0.07331617176532745, + "learning_rate": 5.229477685611799e-07, + "loss": 0.009, + "step": 11040 + }, + { + "epoch": 18.509212730318257, + "grad_norm": 0.05208346247673035, + "learning_rate": 5.121830295164182e-07, + "loss": 0.0102, + "step": 11050 + }, + { + "epoch": 18.525963149078727, + "grad_norm": 0.05402592569589615, + "learning_rate": 5.015283158114853e-07, + "loss": 0.012, + "step": 11060 + }, + { + "epoch": 18.542713567839197, + "grad_norm": 0.05138329043984413, + "learning_rate": 4.90983708361849e-07, + "loss": 0.0092, + "step": 11070 + }, + { + "epoch": 18.559463986599663, + "grad_norm": 0.07261831313371658, + "learning_rate": 4.805492872467942e-07, + "loss": 0.0111, + "step": 11080 + }, + { + "epoch": 18.576214405360133, + "grad_norm": 0.04949916899204254, + "learning_rate": 4.702251317088152e-07, + "loss": 0.0105, + "step": 11090 + }, + { + "epoch": 18.592964824120603, + "grad_norm": 0.0791100412607193, + "learning_rate": 4.600113201530093e-07, + "loss": 0.0105, + "step": 11100 + }, + { + "epoch": 18.609715242881073, + "grad_norm": 0.07842051982879639, + "learning_rate": 4.499079301464859e-07, + "loss": 0.0098, + "step": 11110 + }, + { + "epoch": 18.62646566164154, + "grad_norm": 0.060268741101026535, + "learning_rate": 4.399150384177786e-07, + "loss": 0.0107, + "step": 11120 + }, + { + "epoch": 18.64321608040201, + "grad_norm": 0.07312291860580444, + "learning_rate": 4.3003272085625376e-07, + "loss": 0.0108, + "step": 11130 + }, + { + "epoch": 18.65996649916248, + "grad_norm": 0.07874402403831482, + "learning_rate": 4.20261052511548e-07, + "loss": 0.0107, + "step": 11140 + }, + { + "epoch": 18.67671691792295, + "grad_norm": 0.0655098631978035, + "learning_rate": 4.106001075929883e-07, + "loss": 0.0096, + "step": 11150 + }, + { + "epoch": 18.693467336683415, + "grad_norm": 0.06539636105298996, + "learning_rate": 4.0104995946902424e-07, + "loss": 0.0092, + "step": 11160 + }, + { + "epoch": 18.710217755443885, + "grad_norm": 0.06471475958824158, + "learning_rate": 3.916106806666869e-07, + "loss": 0.0114, + "step": 11170 + }, + { + "epoch": 18.726968174204355, + "grad_norm": 0.047799427062273026, + "learning_rate": 3.8228234287102423e-07, + "loss": 0.0103, + "step": 11180 + }, + { + "epoch": 18.743718592964825, + "grad_norm": 0.05634230747818947, + "learning_rate": 3.730650169245597e-07, + "loss": 0.0102, + "step": 11190 + }, + { + "epoch": 18.76046901172529, + "grad_norm": 0.059011902660131454, + "learning_rate": 3.639587728267579e-07, + "loss": 0.0119, + "step": 11200 + }, + { + "epoch": 18.77721943048576, + "grad_norm": 0.0783902257680893, + "learning_rate": 3.549636797334915e-07, + "loss": 0.013, + "step": 11210 + }, + { + "epoch": 18.79396984924623, + "grad_norm": 0.061978306621313095, + "learning_rate": 3.460798059565101e-07, + "loss": 0.01, + "step": 11220 + }, + { + "epoch": 18.8107202680067, + "grad_norm": 0.07768058031797409, + "learning_rate": 3.3730721896293227e-07, + "loss": 0.0124, + "step": 11230 + }, + { + "epoch": 18.827470686767168, + "grad_norm": 0.07385557144880295, + "learning_rate": 3.286459853747259e-07, + "loss": 0.0108, + "step": 11240 + }, + { + "epoch": 18.844221105527637, + "grad_norm": 0.08369773626327515, + "learning_rate": 3.200961709681988e-07, + "loss": 0.0124, + "step": 11250 + }, + { + "epoch": 18.860971524288107, + "grad_norm": 0.04872307553887367, + "learning_rate": 3.1165784067351213e-07, + "loss": 0.0095, + "step": 11260 + }, + { + "epoch": 18.877721943048577, + "grad_norm": 0.0770203247666359, + "learning_rate": 3.033310585741761e-07, + "loss": 0.0107, + "step": 11270 + }, + { + "epoch": 18.894472361809044, + "grad_norm": 0.07913050055503845, + "learning_rate": 2.951158879065652e-07, + "loss": 0.009, + "step": 11280 + }, + { + "epoch": 18.911222780569513, + "grad_norm": 0.05107510834932327, + "learning_rate": 2.870123910594369e-07, + "loss": 0.0101, + "step": 11290 + }, + { + "epoch": 18.927973199329983, + "grad_norm": 0.08134544640779495, + "learning_rate": 2.790206295734654e-07, + "loss": 0.0087, + "step": 11300 + }, + { + "epoch": 18.944723618090453, + "grad_norm": 0.06199859082698822, + "learning_rate": 2.711406641407688e-07, + "loss": 0.009, + "step": 11310 + }, + { + "epoch": 18.96147403685092, + "grad_norm": 0.10949202626943588, + "learning_rate": 2.633725546044441e-07, + "loss": 0.0091, + "step": 11320 + }, + { + "epoch": 18.97822445561139, + "grad_norm": 0.053913816809654236, + "learning_rate": 2.5571635995812136e-07, + "loss": 0.0105, + "step": 11330 + }, + { + "epoch": 18.99497487437186, + "grad_norm": 0.06924161314964294, + "learning_rate": 2.4817213834551046e-07, + "loss": 0.0094, + "step": 11340 + }, + { + "epoch": 19.01172529313233, + "grad_norm": 0.08277201652526855, + "learning_rate": 2.4073994705995637e-07, + "loss": 0.0118, + "step": 11350 + }, + { + "epoch": 19.028475711892796, + "grad_norm": 0.08757414668798447, + "learning_rate": 2.3341984254401648e-07, + "loss": 0.0116, + "step": 11360 + }, + { + "epoch": 19.045226130653266, + "grad_norm": 0.07003695517778397, + "learning_rate": 2.2621188038901563e-07, + "loss": 0.0108, + "step": 11370 + }, + { + "epoch": 19.061976549413735, + "grad_norm": 0.07919664680957794, + "learning_rate": 2.191161153346366e-07, + "loss": 0.0103, + "step": 11380 + }, + { + "epoch": 19.078726968174205, + "grad_norm": 0.07476326823234558, + "learning_rate": 2.121326012684971e-07, + "loss": 0.0113, + "step": 11390 + }, + { + "epoch": 19.09547738693467, + "grad_norm": 0.04848461598157883, + "learning_rate": 2.052613912257467e-07, + "loss": 0.0104, + "step": 11400 + }, + { + "epoch": 19.11222780569514, + "grad_norm": 0.07294335961341858, + "learning_rate": 1.9850253738865564e-07, + "loss": 0.0116, + "step": 11410 + }, + { + "epoch": 19.12897822445561, + "grad_norm": 0.07061304897069931, + "learning_rate": 1.9185609108622825e-07, + "loss": 0.01, + "step": 11420 + }, + { + "epoch": 19.14572864321608, + "grad_norm": 0.047857094556093216, + "learning_rate": 1.853221027938068e-07, + "loss": 0.0089, + "step": 11430 + }, + { + "epoch": 19.162479061976548, + "grad_norm": 0.07067311555147171, + "learning_rate": 1.7890062213268665e-07, + "loss": 0.0111, + "step": 11440 + }, + { + "epoch": 19.179229480737018, + "grad_norm": 0.08010688424110413, + "learning_rate": 1.7259169786974672e-07, + "loss": 0.0105, + "step": 11450 + }, + { + "epoch": 19.195979899497488, + "grad_norm": 0.06564082950353622, + "learning_rate": 1.663953779170746e-07, + "loss": 0.0113, + "step": 11460 + }, + { + "epoch": 19.212730318257957, + "grad_norm": 0.055230334401130676, + "learning_rate": 1.6031170933159867e-07, + "loss": 0.0101, + "step": 11470 + }, + { + "epoch": 19.229480737018424, + "grad_norm": 0.06035526841878891, + "learning_rate": 1.5434073831473994e-07, + "loss": 0.009, + "step": 11480 + }, + { + "epoch": 19.246231155778894, + "grad_norm": 0.06482507288455963, + "learning_rate": 1.4848251021205405e-07, + "loss": 0.0086, + "step": 11490 + }, + { + "epoch": 19.262981574539364, + "grad_norm": 0.041663289070129395, + "learning_rate": 1.4273706951288989e-07, + "loss": 0.0103, + "step": 11500 + }, + { + "epoch": 19.279731993299833, + "grad_norm": 0.061265233904123306, + "learning_rate": 1.3710445985004815e-07, + "loss": 0.0112, + "step": 11510 + }, + { + "epoch": 19.2964824120603, + "grad_norm": 0.05612942576408386, + "learning_rate": 1.3158472399945996e-07, + "loss": 0.0103, + "step": 11520 + }, + { + "epoch": 19.31323283082077, + "grad_norm": 0.05675517022609711, + "learning_rate": 1.2617790387984385e-07, + "loss": 0.0128, + "step": 11530 + }, + { + "epoch": 19.32998324958124, + "grad_norm": 0.07928426563739777, + "learning_rate": 1.2088404055241088e-07, + "loss": 0.0111, + "step": 11540 + }, + { + "epoch": 19.34673366834171, + "grad_norm": 0.05637091398239136, + "learning_rate": 1.1570317422053e-07, + "loss": 0.0095, + "step": 11550 + }, + { + "epoch": 19.363484087102176, + "grad_norm": 0.08174281567335129, + "learning_rate": 1.1063534422943833e-07, + "loss": 0.0117, + "step": 11560 + }, + { + "epoch": 19.380234505862646, + "grad_norm": 0.09327259659767151, + "learning_rate": 1.0568058906593791e-07, + "loss": 0.0091, + "step": 11570 + }, + { + "epoch": 19.396984924623116, + "grad_norm": 0.04892966151237488, + "learning_rate": 1.0083894635809943e-07, + "loss": 0.0104, + "step": 11580 + }, + { + "epoch": 19.413735343383586, + "grad_norm": 0.08754295855760574, + "learning_rate": 9.611045287497899e-08, + "loss": 0.0121, + "step": 11590 + }, + { + "epoch": 19.430485762144052, + "grad_norm": 0.061934325844049454, + "learning_rate": 9.149514452634511e-08, + "loss": 0.0102, + "step": 11600 + }, + { + "epoch": 19.44723618090452, + "grad_norm": 0.0838724821805954, + "learning_rate": 8.699305636239386e-08, + "loss": 0.012, + "step": 11610 + }, + { + "epoch": 19.46398659966499, + "grad_norm": 0.07911603897809982, + "learning_rate": 8.260422257349409e-08, + "loss": 0.0133, + "step": 11620 + }, + { + "epoch": 19.48073701842546, + "grad_norm": 0.09958657622337341, + "learning_rate": 7.832867648992104e-08, + "loss": 0.011, + "step": 11630 + }, + { + "epoch": 19.497487437185928, + "grad_norm": 0.08941280841827393, + "learning_rate": 7.416645058160642e-08, + "loss": 0.0115, + "step": 11640 + }, + { + "epoch": 19.514237855946398, + "grad_norm": 0.06926604360342026, + "learning_rate": 7.011757645788541e-08, + "loss": 0.0115, + "step": 11650 + }, + { + "epoch": 19.530988274706868, + "grad_norm": 0.06882479041814804, + "learning_rate": 6.618208486727006e-08, + "loss": 0.0094, + "step": 11660 + }, + { + "epoch": 19.547738693467338, + "grad_norm": 0.052707117050886154, + "learning_rate": 6.236000569719957e-08, + "loss": 0.0101, + "step": 11670 + }, + { + "epoch": 19.564489112227804, + "grad_norm": 0.055740345269441605, + "learning_rate": 5.865136797382875e-08, + "loss": 0.0102, + "step": 11680 + }, + { + "epoch": 19.581239530988274, + "grad_norm": 0.04945455119013786, + "learning_rate": 5.505619986179156e-08, + "loss": 0.0087, + "step": 11690 + }, + { + "epoch": 19.597989949748744, + "grad_norm": 0.07951685786247253, + "learning_rate": 5.157452866400292e-08, + "loss": 0.0121, + "step": 11700 + }, + { + "epoch": 19.614740368509214, + "grad_norm": 0.05180259048938751, + "learning_rate": 4.820638082144058e-08, + "loss": 0.0103, + "step": 11710 + }, + { + "epoch": 19.631490787269684, + "grad_norm": 0.03335301950573921, + "learning_rate": 4.495178191294691e-08, + "loss": 0.0094, + "step": 11720 + }, + { + "epoch": 19.64824120603015, + "grad_norm": 0.032290928065776825, + "learning_rate": 4.181075665504075e-08, + "loss": 0.0092, + "step": 11730 + }, + { + "epoch": 19.66499162479062, + "grad_norm": 0.07559007406234741, + "learning_rate": 3.8783328901719203e-08, + "loss": 0.0113, + "step": 11740 + }, + { + "epoch": 19.68174204355109, + "grad_norm": 0.06876268982887268, + "learning_rate": 3.586952164428281e-08, + "loss": 0.0114, + "step": 11750 + }, + { + "epoch": 19.698492462311556, + "grad_norm": 0.06580738723278046, + "learning_rate": 3.3069357011162314e-08, + "loss": 0.0103, + "step": 11760 + }, + { + "epoch": 19.715242881072026, + "grad_norm": 0.06227555125951767, + "learning_rate": 3.038285626774551e-08, + "loss": 0.0105, + "step": 11770 + }, + { + "epoch": 19.731993299832496, + "grad_norm": 0.07759340107440948, + "learning_rate": 2.781003981622232e-08, + "loss": 0.0084, + "step": 11780 + }, + { + "epoch": 19.748743718592966, + "grad_norm": 0.0702289342880249, + "learning_rate": 2.535092719541998e-08, + "loss": 0.0099, + "step": 11790 + }, + { + "epoch": 19.765494137353436, + "grad_norm": 0.07710360735654831, + "learning_rate": 2.3005537080668105e-08, + "loss": 0.0113, + "step": 11800 + }, + { + "epoch": 19.782244556113902, + "grad_norm": 0.09792066365480423, + "learning_rate": 2.0773887283642156e-08, + "loss": 0.0106, + "step": 11810 + }, + { + "epoch": 19.798994974874372, + "grad_norm": 0.04683520272374153, + "learning_rate": 1.865599475224189e-08, + "loss": 0.0097, + "step": 11820 + }, + { + "epoch": 19.81574539363484, + "grad_norm": 0.06832157075405121, + "learning_rate": 1.6651875570451447e-08, + "loss": 0.0124, + "step": 11830 + }, + { + "epoch": 19.832495812395308, + "grad_norm": 0.06539963185787201, + "learning_rate": 1.4761544958222794e-08, + "loss": 0.0117, + "step": 11840 + }, + { + "epoch": 19.849246231155778, + "grad_norm": 0.06669726967811584, + "learning_rate": 1.2985017271364141e-08, + "loss": 0.0119, + "step": 11850 + }, + { + "epoch": 19.865996649916248, + "grad_norm": 0.0661471039056778, + "learning_rate": 1.132230600141837e-08, + "loss": 0.0099, + "step": 11860 + }, + { + "epoch": 19.882747068676718, + "grad_norm": 0.05712989717721939, + "learning_rate": 9.77342377557644e-09, + "loss": 0.0116, + "step": 11870 + }, + { + "epoch": 19.899497487437188, + "grad_norm": 0.05498345196247101, + "learning_rate": 8.338382356565811e-09, + "loss": 0.0091, + "step": 11880 + }, + { + "epoch": 19.916247906197654, + "grad_norm": 0.07772324234247208, + "learning_rate": 7.017192642573833e-09, + "loss": 0.0104, + "step": 11890 + }, + { + "epoch": 19.932998324958124, + "grad_norm": 0.071397565305233, + "learning_rate": 5.809864667157827e-09, + "loss": 0.0113, + "step": 11900 + }, + { + "epoch": 19.949748743718594, + "grad_norm": 0.07486724853515625, + "learning_rate": 4.716407599170136e-09, + "loss": 0.0114, + "step": 11910 + }, + { + "epoch": 19.96649916247906, + "grad_norm": 0.06778570264577866, + "learning_rate": 3.736829742693182e-09, + "loss": 0.0115, + "step": 11920 + }, + { + "epoch": 19.98324958123953, + "grad_norm": 0.0631963312625885, + "learning_rate": 2.8711385369678543e-09, + "loss": 0.0129, + "step": 11930 + }, + { + "epoch": 20.0, + "grad_norm": 0.078452467918396, + "learning_rate": 2.1193405563418867e-09, + "loss": 0.0116, + "step": 11940 + }, + { + "epoch": 20.01675041876047, + "grad_norm": 0.05296291038393974, + "learning_rate": 1.4814415102248903e-09, + "loss": 0.0124, + "step": 11950 + }, + { + "epoch": 20.03350083752094, + "grad_norm": 0.08162922412157059, + "learning_rate": 9.574462430333998e-10, + "loss": 0.0103, + "step": 11960 + }, + { + "epoch": 20.050251256281406, + "grad_norm": 0.06841154396533966, + "learning_rate": 5.473587341642273e-10, + "loss": 0.0124, + "step": 11970 + }, + { + "epoch": 20.067001675041876, + "grad_norm": 0.07906953245401382, + "learning_rate": 2.5118209795782497e-10, + "loss": 0.0101, + "step": 11980 + }, + { + "epoch": 20.083752093802346, + "grad_norm": 0.0642547532916069, + "learning_rate": 6.891858368163195e-11, + "loss": 0.0098, + "step": 11990 + }, + { + "epoch": 20.100502512562816, + "grad_norm": 0.07142816483974457, + "learning_rate": 5.695755050938445e-13, + "loss": 0.0117, + "step": 12000 + }, + { + "epoch": 20.100502512562816, + "step": 12000, + "total_flos": 0.0, + "train_loss": 0.02008304153382778, + "train_runtime": 43219.244, + "train_samples_per_second": 17.77, + "train_steps_per_second": 0.278 + } + ], + "logging_steps": 10, + "max_steps": 12000, + "num_input_tokens_seen": 0, + "num_train_epochs": 21, + "save_steps": 12001, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-5000/config.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-5000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..47d7b77e4255263f3803a4b6f94ca6831e459b70 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-5000/config.json @@ -0,0 +1,64 @@ +{ + "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" + ], + "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", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-5000/experiment_cfg/metadata.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-5000/experiment_cfg/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..23dcfa871a88e56280f945357aef0296bd071711 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-5000/experiment_cfg/metadata.json @@ -0,0 +1,431 @@ +{ + "new_embodiment": { + "statistics": { + "state": { + "base_position": { + "max": [ + 7.778976917266846, + 0.33403685688972473, + 0.7223060727119446 + ], + "min": [ + -4.821934223175049, + -6.785110950469971, + 0.6994727253913879 + ], + "mean": [ + 2.577927505176127, + -2.1525618086425604, + 0.7003541659101092 + ], + "std": [ + 1.6720365948940972, + 1.3768414767482617, + 0.0009102921174316669 + ], + "q01": [ + -0.7605057497562686, + -5.518671947805291, + 0.6999764703640262 + ], + "q99": [ + 6.038879433964306, + -0.5168299981147954, + 0.7029182461269848 + ] + }, + "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, + 0.24782371371751144, + 0.5971093380149053 + ], + "std": [ + 0.0, + 0.0, + 0.6642559499983499, + 0.3751891785184018 + ], + "q01": [ + 0.0, + 0.0, + -1.0, + 4.194193576416142e-06 + ], + "q99": [ + 0.0, + 0.0, + 1.0, + 1.0 + ] + }, + "end_effector_position_relative": { + "max": [ + 0.8848381638526917, + 0.8569459915161133, + 1.053884744644165 + ], + "min": [ + -0.5828549861907959, + -0.8486236333847046, + -0.3446122705936432 + ], + "mean": [ + 0.25006971579930243, + -0.005716176032514759, + 0.4426027721228362 + ], + "std": [ + 0.15838384751435522, + 0.25141504139800513, + 0.20646816234620438 + ], + "q01": [ + -0.017003721077380903, + -0.47738320165276543, + 0.1917563409753314 + ], + "q99": [ + 0.5401029199137674, + 0.4706646609268599, + 0.6965586537529516 + ] + }, + "end_effector_rotation_relative": { + "max": [ + 0.9999949932098389, + 0.9997354745864868, + 0.9900747537612915, + 0.9735639095306396 + ], + "min": [ + -0.9999914169311523, + -0.9999884366989136, + -0.9975369572639465, + 6.754255821306288e-08 + ], + "mean": [ + -0.22531628862537245, + 0.045669797538445966, + -0.06098055586474368, + 0.12292005493282424 + ], + "std": [ + 0.8506341437448487, + 0.2880763358322842, + 0.30816573781739026, + 0.16366165154486362 + ], + "q01": [ + -0.9957510593732032, + -0.5584741016891311, + -0.5371647381159943, + 0.0013784815838340492 + ], + "q99": [ + 0.9899375209218226, + 0.6327878884078038, + 0.5071539024531994, + 0.43555975763914345 + ] + }, + "gripper_qpos": { + "max": [ + 0.05289580300450325, + 0.00937709677964449 + ], + "min": [ + -0.011669117026031017, + -0.06307756155729294 + ], + "mean": [ + 0.03196506153371015, + -0.03229791582989581 + ], + "std": [ + 0.01171640503534049, + 0.01140016755425199 + ], + "q01": [ + 0.006290779996034005, + -0.04059717954551848 + ], + "q99": [ + 0.04038165838477291, + -0.006363076424932449 + ] + } + }, + "action": { + "base_motion": { + "max": [ + 1.0, + 1.0, + 1.0, + 0.0 + ], + "min": [ + -1.0, + -1.0, + -1.0, + 0.0 + ], + "mean": [ + 0.006112990930475949, + 0.002504955933349057, + -0.001639409958610569, + 0.0 + ], + "std": [ + 0.11478077265783658, + 0.1002662368589144, + 0.0807258815139021, + 0.0 + ], + "q01": [ + -0.16969970921613756, + -0.19919537649977875, + -0.15012139866371593, + 0.0 + ], + "q99": [ + 0.2466625453958827, + 0.25040842772679267, + 0.12777701094549537, + 0.0 + ] + }, + "control_mode": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.9253067479599785 + ], + "std": [ + 0.3792397014797737 + ], + "q01": [ + -1.0 + ], + "q99": [ + -0.1974093068624371 + ] + }, + "end_effector_position": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + -0.006754222051360652, + 0.0027854853184013867, + -0.08087572130330477 + ], + "std": [ + 0.45779905420789546, + 0.40578975219522756, + 0.4062694098132682 + ], + "q01": [ + -0.968295960290036, + -0.9240402337368848, + -0.8249359802202183 + ], + "q99": [ + 0.8046634915851651, + 0.9212243025654695, + 0.838433711437593 + ] + }, + "end_effector_rotation": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.008633749251857347, + -0.020406457524911718, + -0.0027639446362883006 + ], + "std": [ + 0.11393972872407104, + 0.13300186268009506, + 0.12192288302920105 + ], + "q01": [ + -0.2668579353776187, + -0.40299251206771697, + -0.34712957892172586 + ], + "q99": [ + 0.32748281264383927, + 0.3179028969292779, + 0.33459405510046736 + ] + }, + "gripper_close": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.3091635769453286 + ], + "std": [ + 0.9510122432876397 + ], + "q01": [ + -1.0 + ], + "q99": [ + 0.7643545615216344 + ] + } + } + }, + "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/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-5000/model.safetensors.index.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-5000/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..5ef5f247b61eb93e89601c7da8334fc087e19750 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-5000/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/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-5000/trainer_state.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-5000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..f527f843e5db5ca29995b97b041e8e10b6605ff3 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-5000/trainer_state.json @@ -0,0 +1,3534 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 8.547008547008547, + "eval_steps": 500, + "global_step": 5000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.017094017094017096, + "grad_norm": 1.0089545249938965, + "learning_rate": 4.5e-07, + "loss": 0.2453, + "step": 10 + }, + { + "epoch": 0.03418803418803419, + "grad_norm": 0.8558177947998047, + "learning_rate": 9.500000000000001e-07, + "loss": 0.2334, + "step": 20 + }, + { + "epoch": 0.05128205128205128, + "grad_norm": 0.7379946112632751, + "learning_rate": 1.45e-06, + "loss": 0.2502, + "step": 30 + }, + { + "epoch": 0.06837606837606838, + "grad_norm": 0.8863121867179871, + "learning_rate": 1.95e-06, + "loss": 0.2143, + "step": 40 + }, + { + "epoch": 0.08547008547008547, + "grad_norm": 1.415384292602539, + "learning_rate": 2.45e-06, + "loss": 0.2502, + "step": 50 + }, + { + "epoch": 0.10256410256410256, + "grad_norm": 0.9582884311676025, + "learning_rate": 2.9499999999999997e-06, + "loss": 0.2322, + "step": 60 + }, + { + "epoch": 0.11965811965811966, + "grad_norm": 0.8592327237129211, + "learning_rate": 3.4500000000000004e-06, + "loss": 0.2117, + "step": 70 + }, + { + "epoch": 0.13675213675213677, + "grad_norm": 0.8079739212989807, + "learning_rate": 3.9499999999999995e-06, + "loss": 0.2001, + "step": 80 + }, + { + "epoch": 0.15384615384615385, + "grad_norm": 0.746914267539978, + "learning_rate": 4.450000000000001e-06, + "loss": 0.2006, + "step": 90 + }, + { + "epoch": 0.17094017094017094, + "grad_norm": 1.141075849533081, + "learning_rate": 4.95e-06, + "loss": 0.1755, + "step": 100 + }, + { + "epoch": 0.18803418803418803, + "grad_norm": 0.9814507365226746, + "learning_rate": 5.45e-06, + "loss": 0.1677, + "step": 110 + }, + { + "epoch": 0.20512820512820512, + "grad_norm": 0.5535752773284912, + "learning_rate": 5.95e-06, + "loss": 0.1627, + "step": 120 + }, + { + "epoch": 0.2222222222222222, + "grad_norm": 0.7861893773078918, + "learning_rate": 6.45e-06, + "loss": 0.1382, + "step": 130 + }, + { + "epoch": 0.23931623931623933, + "grad_norm": 0.43815529346466064, + "learning_rate": 6.95e-06, + "loss": 0.1404, + "step": 140 + }, + { + "epoch": 0.2564102564102564, + "grad_norm": 0.6113188862800598, + "learning_rate": 7.45e-06, + "loss": 0.1303, + "step": 150 + }, + { + "epoch": 0.27350427350427353, + "grad_norm": 0.4023075997829437, + "learning_rate": 7.95e-06, + "loss": 0.1121, + "step": 160 + }, + { + "epoch": 0.2905982905982906, + "grad_norm": 0.43074819445610046, + "learning_rate": 8.45e-06, + "loss": 0.1078, + "step": 170 + }, + { + "epoch": 0.3076923076923077, + "grad_norm": 0.28252357244491577, + "learning_rate": 8.95e-06, + "loss": 0.0936, + "step": 180 + }, + { + "epoch": 0.3247863247863248, + "grad_norm": 0.23254995048046112, + "learning_rate": 9.450000000000001e-06, + "loss": 0.0877, + "step": 190 + }, + { + "epoch": 0.3418803418803419, + "grad_norm": 0.20975816249847412, + "learning_rate": 9.95e-06, + "loss": 0.083, + "step": 200 + }, + { + "epoch": 0.358974358974359, + "grad_norm": 0.27841970324516296, + "learning_rate": 1.045e-05, + "loss": 0.0863, + "step": 210 + }, + { + "epoch": 0.37606837606837606, + "grad_norm": 0.2387871891260147, + "learning_rate": 1.095e-05, + "loss": 0.0822, + "step": 220 + }, + { + "epoch": 0.39316239316239315, + "grad_norm": 0.34779876470565796, + "learning_rate": 1.145e-05, + "loss": 0.0802, + "step": 230 + }, + { + "epoch": 0.41025641025641024, + "grad_norm": 0.2469312846660614, + "learning_rate": 1.1949999999999999e-05, + "loss": 0.0801, + "step": 240 + }, + { + "epoch": 0.42735042735042733, + "grad_norm": 0.18633294105529785, + "learning_rate": 1.245e-05, + "loss": 0.0736, + "step": 250 + }, + { + "epoch": 0.4444444444444444, + "grad_norm": 0.18194709718227386, + "learning_rate": 1.295e-05, + "loss": 0.0722, + "step": 260 + }, + { + "epoch": 0.46153846153846156, + "grad_norm": 0.21349109709262848, + "learning_rate": 1.345e-05, + "loss": 0.0732, + "step": 270 + }, + { + "epoch": 0.47863247863247865, + "grad_norm": 0.17933063209056854, + "learning_rate": 1.395e-05, + "loss": 0.0734, + "step": 280 + }, + { + "epoch": 0.49572649572649574, + "grad_norm": 0.18863998353481293, + "learning_rate": 1.445e-05, + "loss": 0.0673, + "step": 290 + }, + { + "epoch": 0.5128205128205128, + "grad_norm": 0.22107668220996857, + "learning_rate": 1.4950000000000001e-05, + "loss": 0.0701, + "step": 300 + }, + { + "epoch": 0.5299145299145299, + "grad_norm": 0.18152156472206116, + "learning_rate": 1.545e-05, + "loss": 0.0665, + "step": 310 + }, + { + "epoch": 0.5470085470085471, + "grad_norm": 0.1960478127002716, + "learning_rate": 1.5949999999999998e-05, + "loss": 0.0636, + "step": 320 + }, + { + "epoch": 0.5641025641025641, + "grad_norm": 0.21717658638954163, + "learning_rate": 1.645e-05, + "loss": 0.0644, + "step": 330 + }, + { + "epoch": 0.5811965811965812, + "grad_norm": 0.21584761142730713, + "learning_rate": 1.695e-05, + "loss": 0.0707, + "step": 340 + }, + { + "epoch": 0.5982905982905983, + "grad_norm": 0.22434745728969574, + "learning_rate": 1.745e-05, + "loss": 0.0651, + "step": 350 + }, + { + "epoch": 0.6153846153846154, + "grad_norm": 0.22187460958957672, + "learning_rate": 1.7950000000000003e-05, + "loss": 0.0636, + "step": 360 + }, + { + "epoch": 0.6324786324786325, + "grad_norm": 0.2252669632434845, + "learning_rate": 1.845e-05, + "loss": 0.0602, + "step": 370 + }, + { + "epoch": 0.6495726495726496, + "grad_norm": 0.1939157098531723, + "learning_rate": 1.8950000000000003e-05, + "loss": 0.0581, + "step": 380 + }, + { + "epoch": 0.6666666666666666, + "grad_norm": 0.19263336062431335, + "learning_rate": 1.945e-05, + "loss": 0.058, + "step": 390 + }, + { + "epoch": 0.6837606837606838, + "grad_norm": 0.1859099417924881, + "learning_rate": 1.995e-05, + "loss": 0.0646, + "step": 400 + }, + { + "epoch": 0.7008547008547008, + "grad_norm": 0.16536593437194824, + "learning_rate": 2.045e-05, + "loss": 0.0571, + "step": 410 + }, + { + "epoch": 0.717948717948718, + "grad_norm": 0.18150511384010315, + "learning_rate": 2.095e-05, + "loss": 0.0604, + "step": 420 + }, + { + "epoch": 0.7350427350427351, + "grad_norm": 0.19648237526416779, + "learning_rate": 2.145e-05, + "loss": 0.0607, + "step": 430 + }, + { + "epoch": 0.7521367521367521, + "grad_norm": 0.19251717627048492, + "learning_rate": 2.195e-05, + "loss": 0.0576, + "step": 440 + }, + { + "epoch": 0.7692307692307693, + "grad_norm": 0.173828586935997, + "learning_rate": 2.245e-05, + "loss": 0.0526, + "step": 450 + }, + { + "epoch": 0.7863247863247863, + "grad_norm": 0.22500741481781006, + "learning_rate": 2.2950000000000002e-05, + "loss": 0.057, + "step": 460 + }, + { + "epoch": 0.8034188034188035, + "grad_norm": 0.20360895991325378, + "learning_rate": 2.345e-05, + "loss": 0.0563, + "step": 470 + }, + { + "epoch": 0.8205128205128205, + "grad_norm": 0.1981644481420517, + "learning_rate": 2.395e-05, + "loss": 0.0498, + "step": 480 + }, + { + "epoch": 0.8376068376068376, + "grad_norm": 0.18252193927764893, + "learning_rate": 2.4449999999999998e-05, + "loss": 0.0532, + "step": 490 + }, + { + "epoch": 0.8547008547008547, + "grad_norm": 0.24334463477134705, + "learning_rate": 2.495e-05, + "loss": 0.054, + "step": 500 + }, + { + "epoch": 0.8717948717948718, + "grad_norm": 0.20569105446338654, + "learning_rate": 2.5450000000000002e-05, + "loss": 0.0474, + "step": 510 + }, + { + "epoch": 0.8888888888888888, + "grad_norm": 0.16904938220977783, + "learning_rate": 2.595e-05, + "loss": 0.0527, + "step": 520 + }, + { + "epoch": 0.905982905982906, + "grad_norm": 0.1478479504585266, + "learning_rate": 2.6450000000000003e-05, + "loss": 0.0494, + "step": 530 + }, + { + "epoch": 0.9230769230769231, + "grad_norm": 0.18176110088825226, + "learning_rate": 2.695e-05, + "loss": 0.0486, + "step": 540 + }, + { + "epoch": 0.9401709401709402, + "grad_norm": 0.19104203581809998, + "learning_rate": 2.7450000000000003e-05, + "loss": 0.0447, + "step": 550 + }, + { + "epoch": 0.9572649572649573, + "grad_norm": 0.21148279309272766, + "learning_rate": 2.795e-05, + "loss": 0.045, + "step": 560 + }, + { + "epoch": 0.9743589743589743, + "grad_norm": 0.1891864538192749, + "learning_rate": 2.845e-05, + "loss": 0.049, + "step": 570 + }, + { + "epoch": 0.9914529914529915, + "grad_norm": 0.18104152381420135, + "learning_rate": 2.895e-05, + "loss": 0.0461, + "step": 580 + }, + { + "epoch": 1.0085470085470085, + "grad_norm": 0.244981586933136, + "learning_rate": 2.945e-05, + "loss": 0.044, + "step": 590 + }, + { + "epoch": 1.0256410256410255, + "grad_norm": 0.1661752462387085, + "learning_rate": 2.995e-05, + "loss": 0.0464, + "step": 600 + }, + { + "epoch": 1.0427350427350428, + "grad_norm": 0.19394716620445251, + "learning_rate": 2.9999953864407506e-05, + "loss": 0.0434, + "step": 610 + }, + { + "epoch": 1.0598290598290598, + "grad_norm": 0.15391552448272705, + "learning_rate": 2.99997943837114e-05, + "loss": 0.0429, + "step": 620 + }, + { + "epoch": 1.0769230769230769, + "grad_norm": 0.17904387414455414, + "learning_rate": 2.9999520989547326e-05, + "loss": 0.0443, + "step": 630 + }, + { + "epoch": 1.0940170940170941, + "grad_norm": 0.1621219515800476, + "learning_rate": 2.9999133683991525e-05, + "loss": 0.0423, + "step": 640 + }, + { + "epoch": 1.1111111111111112, + "grad_norm": 0.1549968719482422, + "learning_rate": 2.9998632469985326e-05, + "loss": 0.0439, + "step": 650 + }, + { + "epoch": 1.1282051282051282, + "grad_norm": 0.17623211443424225, + "learning_rate": 2.999801735133512e-05, + "loss": 0.0439, + "step": 660 + }, + { + "epoch": 1.1452991452991452, + "grad_norm": 0.20380640029907227, + "learning_rate": 2.9997288332712325e-05, + "loss": 0.0424, + "step": 670 + }, + { + "epoch": 1.1623931623931625, + "grad_norm": 0.15982045233249664, + "learning_rate": 2.999644541965335e-05, + "loss": 0.0409, + "step": 680 + }, + { + "epoch": 1.1794871794871795, + "grad_norm": 0.16698749363422394, + "learning_rate": 2.9995488618559554e-05, + "loss": 0.0376, + "step": 690 + }, + { + "epoch": 1.1965811965811965, + "grad_norm": 0.13415130972862244, + "learning_rate": 2.9994417936697218e-05, + "loss": 0.0412, + "step": 700 + }, + { + "epoch": 1.2136752136752136, + "grad_norm": 0.13148538768291473, + "learning_rate": 2.9993233382197454e-05, + "loss": 0.0376, + "step": 710 + }, + { + "epoch": 1.2307692307692308, + "grad_norm": 0.16429367661476135, + "learning_rate": 2.999193496405616e-05, + "loss": 0.0398, + "step": 720 + }, + { + "epoch": 1.2478632478632479, + "grad_norm": 0.15812724828720093, + "learning_rate": 2.9990522692133962e-05, + "loss": 0.0406, + "step": 730 + }, + { + "epoch": 1.264957264957265, + "grad_norm": 0.13753320276737213, + "learning_rate": 2.9988996577156134e-05, + "loss": 0.0402, + "step": 740 + }, + { + "epoch": 1.282051282051282, + "grad_norm": 0.139229878783226, + "learning_rate": 2.9987356630712496e-05, + "loss": 0.0387, + "step": 750 + }, + { + "epoch": 1.2991452991452992, + "grad_norm": 0.15705733001232147, + "learning_rate": 2.9985602865257354e-05, + "loss": 0.0389, + "step": 760 + }, + { + "epoch": 1.3162393162393162, + "grad_norm": 0.14831461012363434, + "learning_rate": 2.998373529410939e-05, + "loss": 0.0402, + "step": 770 + }, + { + "epoch": 1.3333333333333333, + "grad_norm": 0.13418765366077423, + "learning_rate": 2.9981753931451567e-05, + "loss": 0.0385, + "step": 780 + }, + { + "epoch": 1.3504273504273505, + "grad_norm": 0.16108068823814392, + "learning_rate": 2.9979658792331016e-05, + "loss": 0.0359, + "step": 790 + }, + { + "epoch": 1.3675213675213675, + "grad_norm": 0.14137201011180878, + "learning_rate": 2.997744989265893e-05, + "loss": 0.0367, + "step": 800 + }, + { + "epoch": 1.3846153846153846, + "grad_norm": 0.15811926126480103, + "learning_rate": 2.9975127249210426e-05, + "loss": 0.0348, + "step": 810 + }, + { + "epoch": 1.4017094017094016, + "grad_norm": 0.1296013742685318, + "learning_rate": 2.9972690879624445e-05, + "loss": 0.034, + "step": 820 + }, + { + "epoch": 1.4188034188034189, + "grad_norm": 0.16444191336631775, + "learning_rate": 2.9970140802403587e-05, + "loss": 0.037, + "step": 830 + }, + { + "epoch": 1.435897435897436, + "grad_norm": 0.1809467077255249, + "learning_rate": 2.9967477036914002e-05, + "loss": 0.037, + "step": 840 + }, + { + "epoch": 1.452991452991453, + "grad_norm": 0.11857502162456512, + "learning_rate": 2.996469960338521e-05, + "loss": 0.0356, + "step": 850 + }, + { + "epoch": 1.4700854700854702, + "grad_norm": 0.12802129983901978, + "learning_rate": 2.9961808522909976e-05, + "loss": 0.0361, + "step": 860 + }, + { + "epoch": 1.4871794871794872, + "grad_norm": 0.1897210031747818, + "learning_rate": 2.995880381744414e-05, + "loss": 0.0356, + "step": 870 + }, + { + "epoch": 1.5042735042735043, + "grad_norm": 0.1568431556224823, + "learning_rate": 2.995568550980643e-05, + "loss": 0.0368, + "step": 880 + }, + { + "epoch": 1.5213675213675213, + "grad_norm": 0.14498373866081238, + "learning_rate": 2.9952453623678334e-05, + "loss": 0.0369, + "step": 890 + }, + { + "epoch": 1.5384615384615383, + "grad_norm": 0.14627397060394287, + "learning_rate": 2.9949108183603867e-05, + "loss": 0.0356, + "step": 900 + }, + { + "epoch": 1.5555555555555556, + "grad_norm": 0.143493190407753, + "learning_rate": 2.994564921498943e-05, + "loss": 0.0341, + "step": 910 + }, + { + "epoch": 1.5726495726495726, + "grad_norm": 0.1655547320842743, + "learning_rate": 2.9942076744103594e-05, + "loss": 0.0322, + "step": 920 + }, + { + "epoch": 1.5897435897435899, + "grad_norm": 0.102423295378685, + "learning_rate": 2.9938390798076886e-05, + "loss": 0.0346, + "step": 930 + }, + { + "epoch": 1.606837606837607, + "grad_norm": 0.1399497091770172, + "learning_rate": 2.993459140490163e-05, + "loss": 0.0332, + "step": 940 + }, + { + "epoch": 1.623931623931624, + "grad_norm": 0.15969839692115784, + "learning_rate": 2.9930678593431684e-05, + "loss": 0.0318, + "step": 950 + }, + { + "epoch": 1.641025641025641, + "grad_norm": 0.16585025191307068, + "learning_rate": 2.9926652393382258e-05, + "loss": 0.0326, + "step": 960 + }, + { + "epoch": 1.658119658119658, + "grad_norm": 0.13431431353092194, + "learning_rate": 2.9922512835329655e-05, + "loss": 0.0344, + "step": 970 + }, + { + "epoch": 1.6752136752136753, + "grad_norm": 0.1281099021434784, + "learning_rate": 2.9918259950711073e-05, + "loss": 0.0312, + "step": 980 + }, + { + "epoch": 1.6923076923076923, + "grad_norm": 0.13682444393634796, + "learning_rate": 2.991389377182435e-05, + "loss": 0.0338, + "step": 990 + }, + { + "epoch": 1.7094017094017095, + "grad_norm": 0.17258574068546295, + "learning_rate": 2.99094143318277e-05, + "loss": 0.03, + "step": 1000 + }, + { + "epoch": 1.7264957264957266, + "grad_norm": 0.12355158478021622, + "learning_rate": 2.9904821664739493e-05, + "loss": 0.0325, + "step": 1010 + }, + { + "epoch": 1.7435897435897436, + "grad_norm": 0.14875563979148865, + "learning_rate": 2.990011580543799e-05, + "loss": 0.0353, + "step": 1020 + }, + { + "epoch": 1.7606837606837606, + "grad_norm": 0.1353384256362915, + "learning_rate": 2.9895296789661065e-05, + "loss": 0.0294, + "step": 1030 + }, + { + "epoch": 1.7777777777777777, + "grad_norm": 0.1349417269229889, + "learning_rate": 2.989036465400594e-05, + "loss": 0.0299, + "step": 1040 + }, + { + "epoch": 1.7948717948717947, + "grad_norm": 0.17602555453777313, + "learning_rate": 2.9885319435928896e-05, + "loss": 0.028, + "step": 1050 + }, + { + "epoch": 1.811965811965812, + "grad_norm": 0.10533279925584793, + "learning_rate": 2.9880161173745017e-05, + "loss": 0.0272, + "step": 1060 + }, + { + "epoch": 1.8290598290598292, + "grad_norm": 0.12225744873285294, + "learning_rate": 2.9874889906627882e-05, + "loss": 0.0319, + "step": 1070 + }, + { + "epoch": 1.8461538461538463, + "grad_norm": 0.1326926201581955, + "learning_rate": 2.9869505674609253e-05, + "loss": 0.0281, + "step": 1080 + }, + { + "epoch": 1.8632478632478633, + "grad_norm": 0.157648503780365, + "learning_rate": 2.9864008518578794e-05, + "loss": 0.0296, + "step": 1090 + }, + { + "epoch": 1.8803418803418803, + "grad_norm": 0.191550612449646, + "learning_rate": 2.9858398480283758e-05, + "loss": 0.0311, + "step": 1100 + }, + { + "epoch": 1.8974358974358974, + "grad_norm": 0.14470480382442474, + "learning_rate": 2.9852675602328644e-05, + "loss": 0.0297, + "step": 1110 + }, + { + "epoch": 1.9145299145299144, + "grad_norm": 0.10989037901163101, + "learning_rate": 2.9846839928174923e-05, + "loss": 0.0279, + "step": 1120 + }, + { + "epoch": 1.9316239316239316, + "grad_norm": 0.12898847460746765, + "learning_rate": 2.984089150214065e-05, + "loss": 0.0324, + "step": 1130 + }, + { + "epoch": 1.9487179487179487, + "grad_norm": 0.1219201385974884, + "learning_rate": 2.9834830369400165e-05, + "loss": 0.0295, + "step": 1140 + }, + { + "epoch": 1.965811965811966, + "grad_norm": 0.14238029718399048, + "learning_rate": 2.9828656575983747e-05, + "loss": 0.0269, + "step": 1150 + }, + { + "epoch": 1.982905982905983, + "grad_norm": 0.12812891602516174, + "learning_rate": 2.9822370168777253e-05, + "loss": 0.0296, + "step": 1160 + }, + { + "epoch": 2.0, + "grad_norm": 0.16687121987342834, + "learning_rate": 2.981597119552177e-05, + "loss": 0.0319, + "step": 1170 + }, + { + "epoch": 2.017094017094017, + "grad_norm": 0.13151830434799194, + "learning_rate": 2.9809459704813236e-05, + "loss": 0.0281, + "step": 1180 + }, + { + "epoch": 2.034188034188034, + "grad_norm": 0.10349870473146439, + "learning_rate": 2.980283574610211e-05, + "loss": 0.026, + "step": 1190 + }, + { + "epoch": 2.051282051282051, + "grad_norm": 0.13649572432041168, + "learning_rate": 2.9796099369692946e-05, + "loss": 0.03, + "step": 1200 + }, + { + "epoch": 2.0683760683760686, + "grad_norm": 0.13860812783241272, + "learning_rate": 2.978925062674404e-05, + "loss": 0.0269, + "step": 1210 + }, + { + "epoch": 2.0854700854700856, + "grad_norm": 0.13259582221508026, + "learning_rate": 2.9782289569267057e-05, + "loss": 0.0275, + "step": 1220 + }, + { + "epoch": 2.1025641025641026, + "grad_norm": 0.15669459104537964, + "learning_rate": 2.977521625012659e-05, + "loss": 0.0307, + "step": 1230 + }, + { + "epoch": 2.1196581196581197, + "grad_norm": 0.1643408089876175, + "learning_rate": 2.9768030723039798e-05, + "loss": 0.0288, + "step": 1240 + }, + { + "epoch": 2.1367521367521367, + "grad_norm": 0.12998151779174805, + "learning_rate": 2.9760733042575984e-05, + "loss": 0.0282, + "step": 1250 + }, + { + "epoch": 2.1538461538461537, + "grad_norm": 0.11236439645290375, + "learning_rate": 2.9753323264156184e-05, + "loss": 0.0276, + "step": 1260 + }, + { + "epoch": 2.1709401709401708, + "grad_norm": 0.13808509707450867, + "learning_rate": 2.974580144405273e-05, + "loss": 0.0274, + "step": 1270 + }, + { + "epoch": 2.1880341880341883, + "grad_norm": 0.13739660382270813, + "learning_rate": 2.973816763938886e-05, + "loss": 0.0283, + "step": 1280 + }, + { + "epoch": 2.2051282051282053, + "grad_norm": 0.13221029937267303, + "learning_rate": 2.9730421908138236e-05, + "loss": 0.0279, + "step": 1290 + }, + { + "epoch": 2.2222222222222223, + "grad_norm": 0.1287679374217987, + "learning_rate": 2.9722564309124546e-05, + "loss": 0.029, + "step": 1300 + }, + { + "epoch": 2.2393162393162394, + "grad_norm": 0.12047926336526871, + "learning_rate": 2.971459490202102e-05, + "loss": 0.0268, + "step": 1310 + }, + { + "epoch": 2.2564102564102564, + "grad_norm": 0.09681694209575653, + "learning_rate": 2.9706513747350017e-05, + "loss": 0.0254, + "step": 1320 + }, + { + "epoch": 2.2735042735042734, + "grad_norm": 0.14194591343402863, + "learning_rate": 2.9698320906482535e-05, + "loss": 0.0295, + "step": 1330 + }, + { + "epoch": 2.2905982905982905, + "grad_norm": 0.12847678363323212, + "learning_rate": 2.9690016441637747e-05, + "loss": 0.0269, + "step": 1340 + }, + { + "epoch": 2.3076923076923075, + "grad_norm": 0.15464898943901062, + "learning_rate": 2.9681600415882556e-05, + "loss": 0.0269, + "step": 1350 + }, + { + "epoch": 2.324786324786325, + "grad_norm": 0.12396308779716492, + "learning_rate": 2.9673072893131068e-05, + "loss": 0.0286, + "step": 1360 + }, + { + "epoch": 2.341880341880342, + "grad_norm": 0.14894050359725952, + "learning_rate": 2.9664433938144162e-05, + "loss": 0.0288, + "step": 1370 + }, + { + "epoch": 2.358974358974359, + "grad_norm": 0.13231822848320007, + "learning_rate": 2.9655683616528958e-05, + "loss": 0.0266, + "step": 1380 + }, + { + "epoch": 2.376068376068376, + "grad_norm": 0.145207017660141, + "learning_rate": 2.9646821994738322e-05, + "loss": 0.0257, + "step": 1390 + }, + { + "epoch": 2.393162393162393, + "grad_norm": 0.14421381056308746, + "learning_rate": 2.9637849140070393e-05, + "loss": 0.0257, + "step": 1400 + }, + { + "epoch": 2.41025641025641, + "grad_norm": 0.1812671273946762, + "learning_rate": 2.9628765120668033e-05, + "loss": 0.0256, + "step": 1410 + }, + { + "epoch": 2.427350427350427, + "grad_norm": 0.1101057231426239, + "learning_rate": 2.9619570005518336e-05, + "loss": 0.0263, + "step": 1420 + }, + { + "epoch": 2.4444444444444446, + "grad_norm": 0.12395047396421432, + "learning_rate": 2.9610263864452077e-05, + "loss": 0.025, + "step": 1430 + }, + { + "epoch": 2.4615384615384617, + "grad_norm": 0.1621936559677124, + "learning_rate": 2.9600846768143225e-05, + "loss": 0.0261, + "step": 1440 + }, + { + "epoch": 2.4786324786324787, + "grad_norm": 0.13965119421482086, + "learning_rate": 2.959131878810836e-05, + "loss": 0.0251, + "step": 1450 + }, + { + "epoch": 2.4957264957264957, + "grad_norm": 0.11430062353610992, + "learning_rate": 2.9581679996706157e-05, + "loss": 0.0263, + "step": 1460 + }, + { + "epoch": 2.5128205128205128, + "grad_norm": 0.1259123533964157, + "learning_rate": 2.9571930467136833e-05, + "loss": 0.0246, + "step": 1470 + }, + { + "epoch": 2.52991452991453, + "grad_norm": 0.13637065887451172, + "learning_rate": 2.956207027344159e-05, + "loss": 0.0253, + "step": 1480 + }, + { + "epoch": 2.547008547008547, + "grad_norm": 0.1521414965391159, + "learning_rate": 2.9552099490502026e-05, + "loss": 0.0248, + "step": 1490 + }, + { + "epoch": 2.564102564102564, + "grad_norm": 0.17049413919448853, + "learning_rate": 2.9542018194039634e-05, + "loss": 0.0262, + "step": 1500 + }, + { + "epoch": 2.5811965811965814, + "grad_norm": 0.09336340427398682, + "learning_rate": 2.953182646061514e-05, + "loss": 0.0263, + "step": 1510 + }, + { + "epoch": 2.5982905982905984, + "grad_norm": 0.12669335305690765, + "learning_rate": 2.9521524367627994e-05, + "loss": 0.0265, + "step": 1520 + }, + { + "epoch": 2.6153846153846154, + "grad_norm": 0.12848782539367676, + "learning_rate": 2.9511111993315743e-05, + "loss": 0.026, + "step": 1530 + }, + { + "epoch": 2.6324786324786325, + "grad_norm": 0.14347173273563385, + "learning_rate": 2.9500589416753447e-05, + "loss": 0.0244, + "step": 1540 + }, + { + "epoch": 2.6495726495726495, + "grad_norm": 0.10938625037670135, + "learning_rate": 2.9489956717853088e-05, + "loss": 0.0243, + "step": 1550 + }, + { + "epoch": 2.6666666666666665, + "grad_norm": 0.11029652506113052, + "learning_rate": 2.9479213977362935e-05, + "loss": 0.0277, + "step": 1560 + }, + { + "epoch": 2.683760683760684, + "grad_norm": 0.14064879715442657, + "learning_rate": 2.9468361276866973e-05, + "loss": 0.0257, + "step": 1570 + }, + { + "epoch": 2.700854700854701, + "grad_norm": 0.13628490269184113, + "learning_rate": 2.9457398698784243e-05, + "loss": 0.0261, + "step": 1580 + }, + { + "epoch": 2.717948717948718, + "grad_norm": 0.10640981048345566, + "learning_rate": 2.944632632636823e-05, + "loss": 0.0236, + "step": 1590 + }, + { + "epoch": 2.735042735042735, + "grad_norm": 0.10401084274053574, + "learning_rate": 2.9435144243706255e-05, + "loss": 0.0262, + "step": 1600 + }, + { + "epoch": 2.752136752136752, + "grad_norm": 0.11106079071760178, + "learning_rate": 2.9423852535718805e-05, + "loss": 0.0249, + "step": 1610 + }, + { + "epoch": 2.769230769230769, + "grad_norm": 0.18114276230335236, + "learning_rate": 2.9412451288158883e-05, + "loss": 0.024, + "step": 1620 + }, + { + "epoch": 2.786324786324786, + "grad_norm": 0.1303853988647461, + "learning_rate": 2.94009405876114e-05, + "loss": 0.0239, + "step": 1630 + }, + { + "epoch": 2.8034188034188032, + "grad_norm": 0.11971442401409149, + "learning_rate": 2.9389320521492464e-05, + "loss": 0.023, + "step": 1640 + }, + { + "epoch": 2.8205128205128203, + "grad_norm": 0.1276145875453949, + "learning_rate": 2.9377591178048765e-05, + "loss": 0.0237, + "step": 1650 + }, + { + "epoch": 2.8376068376068377, + "grad_norm": 0.1473803073167801, + "learning_rate": 2.9365752646356857e-05, + "loss": 0.0232, + "step": 1660 + }, + { + "epoch": 2.8547008547008548, + "grad_norm": 0.14460980892181396, + "learning_rate": 2.935380501632253e-05, + "loss": 0.0236, + "step": 1670 + }, + { + "epoch": 2.871794871794872, + "grad_norm": 0.1054682806134224, + "learning_rate": 2.934174837868008e-05, + "loss": 0.0237, + "step": 1680 + }, + { + "epoch": 2.888888888888889, + "grad_norm": 0.09902238100767136, + "learning_rate": 2.932958282499167e-05, + "loss": 0.0262, + "step": 1690 + }, + { + "epoch": 2.905982905982906, + "grad_norm": 0.14733590185642242, + "learning_rate": 2.9317308447646585e-05, + "loss": 0.0275, + "step": 1700 + }, + { + "epoch": 2.9230769230769234, + "grad_norm": 0.1391238123178482, + "learning_rate": 2.9304925339860565e-05, + "loss": 0.0254, + "step": 1710 + }, + { + "epoch": 2.9401709401709404, + "grad_norm": 0.1650138944387436, + "learning_rate": 2.929243359567509e-05, + "loss": 0.0228, + "step": 1720 + }, + { + "epoch": 2.9572649572649574, + "grad_norm": 0.13160136342048645, + "learning_rate": 2.9279833309956637e-05, + "loss": 0.0245, + "step": 1730 + }, + { + "epoch": 2.9743589743589745, + "grad_norm": 0.1312543749809265, + "learning_rate": 2.9267124578396013e-05, + "loss": 0.0221, + "step": 1740 + }, + { + "epoch": 2.9914529914529915, + "grad_norm": 0.13742756843566895, + "learning_rate": 2.9254307497507584e-05, + "loss": 0.0256, + "step": 1750 + }, + { + "epoch": 3.0085470085470085, + "grad_norm": 0.08947888761758804, + "learning_rate": 2.924138216462856e-05, + "loss": 0.0247, + "step": 1760 + }, + { + "epoch": 3.0256410256410255, + "grad_norm": 0.09200987219810486, + "learning_rate": 2.922834867791825e-05, + "loss": 0.0223, + "step": 1770 + }, + { + "epoch": 3.0427350427350426, + "grad_norm": 0.14834287762641907, + "learning_rate": 2.9215207136357322e-05, + "loss": 0.0254, + "step": 1780 + }, + { + "epoch": 3.0598290598290596, + "grad_norm": 0.13650742173194885, + "learning_rate": 2.9201957639747056e-05, + "loss": 0.0248, + "step": 1790 + }, + { + "epoch": 3.076923076923077, + "grad_norm": 0.10533406585454941, + "learning_rate": 2.9188600288708565e-05, + "loss": 0.0239, + "step": 1800 + }, + { + "epoch": 3.094017094017094, + "grad_norm": 0.09621796756982803, + "learning_rate": 2.9175135184682053e-05, + "loss": 0.0219, + "step": 1810 + }, + { + "epoch": 3.111111111111111, + "grad_norm": 0.1239803209900856, + "learning_rate": 2.9161562429926038e-05, + "loss": 0.0228, + "step": 1820 + }, + { + "epoch": 3.128205128205128, + "grad_norm": 0.13734465837478638, + "learning_rate": 2.9147882127516568e-05, + "loss": 0.0229, + "step": 1830 + }, + { + "epoch": 3.1452991452991452, + "grad_norm": 0.11425598710775375, + "learning_rate": 2.913409438134644e-05, + "loss": 0.0224, + "step": 1840 + }, + { + "epoch": 3.1623931623931623, + "grad_norm": 0.16756656765937805, + "learning_rate": 2.9120199296124436e-05, + "loss": 0.025, + "step": 1850 + }, + { + "epoch": 3.1794871794871793, + "grad_norm": 0.11103396117687225, + "learning_rate": 2.9106196977374477e-05, + "loss": 0.0236, + "step": 1860 + }, + { + "epoch": 3.1965811965811968, + "grad_norm": 0.10024338215589523, + "learning_rate": 2.9092087531434878e-05, + "loss": 0.0219, + "step": 1870 + }, + { + "epoch": 3.213675213675214, + "grad_norm": 0.11945660412311554, + "learning_rate": 2.9077871065457495e-05, + "loss": 0.0239, + "step": 1880 + }, + { + "epoch": 3.230769230769231, + "grad_norm": 0.11576039344072342, + "learning_rate": 2.9063547687406945e-05, + "loss": 0.0243, + "step": 1890 + }, + { + "epoch": 3.247863247863248, + "grad_norm": 0.09603351354598999, + "learning_rate": 2.9049117506059756e-05, + "loss": 0.0228, + "step": 1900 + }, + { + "epoch": 3.264957264957265, + "grad_norm": 0.07817462831735611, + "learning_rate": 2.9034580631003564e-05, + "loss": 0.0235, + "step": 1910 + }, + { + "epoch": 3.282051282051282, + "grad_norm": 0.11091442406177521, + "learning_rate": 2.9019937172636276e-05, + "loss": 0.0237, + "step": 1920 + }, + { + "epoch": 3.299145299145299, + "grad_norm": 0.13915278017520905, + "learning_rate": 2.9005187242165228e-05, + "loss": 0.0228, + "step": 1930 + }, + { + "epoch": 3.316239316239316, + "grad_norm": 0.09350527077913284, + "learning_rate": 2.899033095160633e-05, + "loss": 0.0231, + "step": 1940 + }, + { + "epoch": 3.3333333333333335, + "grad_norm": 0.11766257882118225, + "learning_rate": 2.8975368413783235e-05, + "loss": 0.0213, + "step": 1950 + }, + { + "epoch": 3.3504273504273505, + "grad_norm": 0.0914580225944519, + "learning_rate": 2.8960299742326473e-05, + "loss": 0.0233, + "step": 1960 + }, + { + "epoch": 3.3675213675213675, + "grad_norm": 0.12332584708929062, + "learning_rate": 2.8945125051672595e-05, + "loss": 0.0245, + "step": 1970 + }, + { + "epoch": 3.3846153846153846, + "grad_norm": 0.10972987115383148, + "learning_rate": 2.892984445706328e-05, + "loss": 0.0234, + "step": 1980 + }, + { + "epoch": 3.4017094017094016, + "grad_norm": 0.1142164021730423, + "learning_rate": 2.891445807454449e-05, + "loss": 0.0235, + "step": 1990 + }, + { + "epoch": 3.4188034188034186, + "grad_norm": 0.13368234038352966, + "learning_rate": 2.8898966020965576e-05, + "loss": 0.0224, + "step": 2000 + }, + { + "epoch": 3.435897435897436, + "grad_norm": 0.10849815607070923, + "learning_rate": 2.8883368413978386e-05, + "loss": 0.0213, + "step": 2010 + }, + { + "epoch": 3.452991452991453, + "grad_norm": 0.09034600108861923, + "learning_rate": 2.8867665372036375e-05, + "loss": 0.0239, + "step": 2020 + }, + { + "epoch": 3.47008547008547, + "grad_norm": 0.12289014458656311, + "learning_rate": 2.8851857014393705e-05, + "loss": 0.022, + "step": 2030 + }, + { + "epoch": 3.4871794871794872, + "grad_norm": 0.15092331171035767, + "learning_rate": 2.8835943461104354e-05, + "loss": 0.0212, + "step": 2040 + }, + { + "epoch": 3.5042735042735043, + "grad_norm": 0.11829588562250137, + "learning_rate": 2.8819924833021176e-05, + "loss": 0.0235, + "step": 2050 + }, + { + "epoch": 3.5213675213675213, + "grad_norm": 0.11106596887111664, + "learning_rate": 2.8803801251795007e-05, + "loss": 0.0223, + "step": 2060 + }, + { + "epoch": 3.5384615384615383, + "grad_norm": 0.0992603749036789, + "learning_rate": 2.8787572839873725e-05, + "loss": 0.024, + "step": 2070 + }, + { + "epoch": 3.5555555555555554, + "grad_norm": 0.11236703395843506, + "learning_rate": 2.8771239720501333e-05, + "loss": 0.0203, + "step": 2080 + }, + { + "epoch": 3.5726495726495724, + "grad_norm": 0.14634686708450317, + "learning_rate": 2.8754802017717015e-05, + "loss": 0.0244, + "step": 2090 + }, + { + "epoch": 3.58974358974359, + "grad_norm": 0.09683041274547577, + "learning_rate": 2.87382598563542e-05, + "loss": 0.0227, + "step": 2100 + }, + { + "epoch": 3.606837606837607, + "grad_norm": 0.13424035906791687, + "learning_rate": 2.8721613362039607e-05, + "loss": 0.0215, + "step": 2110 + }, + { + "epoch": 3.623931623931624, + "grad_norm": 0.10873326659202576, + "learning_rate": 2.8704862661192295e-05, + "loss": 0.0215, + "step": 2120 + }, + { + "epoch": 3.641025641025641, + "grad_norm": 0.0864587277173996, + "learning_rate": 2.86880078810227e-05, + "loss": 0.0251, + "step": 2130 + }, + { + "epoch": 3.658119658119658, + "grad_norm": 0.11012615263462067, + "learning_rate": 2.8671049149531677e-05, + "loss": 0.0199, + "step": 2140 + }, + { + "epoch": 3.6752136752136755, + "grad_norm": 0.18411065638065338, + "learning_rate": 2.8653986595509524e-05, + "loss": 0.0232, + "step": 2150 + }, + { + "epoch": 3.6923076923076925, + "grad_norm": 0.10877490788698196, + "learning_rate": 2.8636820348534987e-05, + "loss": 0.0207, + "step": 2160 + }, + { + "epoch": 3.7094017094017095, + "grad_norm": 0.10099867731332779, + "learning_rate": 2.861955053897431e-05, + "loss": 0.0175, + "step": 2170 + }, + { + "epoch": 3.7264957264957266, + "grad_norm": 0.08751079440116882, + "learning_rate": 2.8602177297980213e-05, + "loss": 0.0212, + "step": 2180 + }, + { + "epoch": 3.7435897435897436, + "grad_norm": 0.13080573081970215, + "learning_rate": 2.8584700757490913e-05, + "loss": 0.0208, + "step": 2190 + }, + { + "epoch": 3.7606837606837606, + "grad_norm": 0.0996410921216011, + "learning_rate": 2.8567121050229122e-05, + "loss": 0.024, + "step": 2200 + }, + { + "epoch": 3.7777777777777777, + "grad_norm": 0.12667258083820343, + "learning_rate": 2.8549438309701028e-05, + "loss": 0.0227, + "step": 2210 + }, + { + "epoch": 3.7948717948717947, + "grad_norm": 0.12759651243686676, + "learning_rate": 2.8531652670195292e-05, + "loss": 0.0215, + "step": 2220 + }, + { + "epoch": 3.8119658119658117, + "grad_norm": 0.10755474865436554, + "learning_rate": 2.8513764266782025e-05, + "loss": 0.0226, + "step": 2230 + }, + { + "epoch": 3.8290598290598292, + "grad_norm": 0.1080416887998581, + "learning_rate": 2.849577323531176e-05, + "loss": 0.021, + "step": 2240 + }, + { + "epoch": 3.8461538461538463, + "grad_norm": 0.09134095907211304, + "learning_rate": 2.8477679712414424e-05, + "loss": 0.0213, + "step": 2250 + }, + { + "epoch": 3.8632478632478633, + "grad_norm": 0.09035451710224152, + "learning_rate": 2.8459483835498298e-05, + "loss": 0.0227, + "step": 2260 + }, + { + "epoch": 3.8803418803418803, + "grad_norm": 0.11327807605266571, + "learning_rate": 2.844118574274896e-05, + "loss": 0.0211, + "step": 2270 + }, + { + "epoch": 3.8974358974358974, + "grad_norm": 0.118441142141819, + "learning_rate": 2.842278557312828e-05, + "loss": 0.0191, + "step": 2280 + }, + { + "epoch": 3.9145299145299144, + "grad_norm": 0.0991198942065239, + "learning_rate": 2.84042834663733e-05, + "loss": 0.0192, + "step": 2290 + }, + { + "epoch": 3.931623931623932, + "grad_norm": 0.08809023350477219, + "learning_rate": 2.8385679562995235e-05, + "loss": 0.0202, + "step": 2300 + }, + { + "epoch": 3.948717948717949, + "grad_norm": 0.10185550898313522, + "learning_rate": 2.8366974004278357e-05, + "loss": 0.0232, + "step": 2310 + }, + { + "epoch": 3.965811965811966, + "grad_norm": 0.11275793612003326, + "learning_rate": 2.834816693227895e-05, + "loss": 0.0197, + "step": 2320 + }, + { + "epoch": 3.982905982905983, + "grad_norm": 0.1458314061164856, + "learning_rate": 2.8329258489824234e-05, + "loss": 0.0246, + "step": 2330 + }, + { + "epoch": 4.0, + "grad_norm": 0.08866702765226364, + "learning_rate": 2.8310248820511248e-05, + "loss": 0.0196, + "step": 2340 + }, + { + "epoch": 4.017094017094017, + "grad_norm": 0.1087232232093811, + "learning_rate": 2.829113806870581e-05, + "loss": 0.0222, + "step": 2350 + }, + { + "epoch": 4.034188034188034, + "grad_norm": 0.13110089302062988, + "learning_rate": 2.8271926379541363e-05, + "loss": 0.0213, + "step": 2360 + }, + { + "epoch": 4.051282051282051, + "grad_norm": 0.07700895518064499, + "learning_rate": 2.8252613898917923e-05, + "loss": 0.02, + "step": 2370 + }, + { + "epoch": 4.068376068376068, + "grad_norm": 0.09605945646762848, + "learning_rate": 2.8233200773500942e-05, + "loss": 0.021, + "step": 2380 + }, + { + "epoch": 4.085470085470085, + "grad_norm": 0.11480522155761719, + "learning_rate": 2.8213687150720204e-05, + "loss": 0.0223, + "step": 2390 + }, + { + "epoch": 4.102564102564102, + "grad_norm": 0.12132875621318817, + "learning_rate": 2.8194073178768703e-05, + "loss": 0.0192, + "step": 2400 + }, + { + "epoch": 4.119658119658119, + "grad_norm": 0.09794069826602936, + "learning_rate": 2.817435900660152e-05, + "loss": 0.0189, + "step": 2410 + }, + { + "epoch": 4.136752136752137, + "grad_norm": 0.09901175647974014, + "learning_rate": 2.8154544783934688e-05, + "loss": 0.0238, + "step": 2420 + }, + { + "epoch": 4.153846153846154, + "grad_norm": 0.10227429866790771, + "learning_rate": 2.8134630661244055e-05, + "loss": 0.0204, + "step": 2430 + }, + { + "epoch": 4.170940170940171, + "grad_norm": 0.12511932849884033, + "learning_rate": 2.8114616789764146e-05, + "loss": 0.0209, + "step": 2440 + }, + { + "epoch": 4.188034188034188, + "grad_norm": 0.10401079803705215, + "learning_rate": 2.8094503321487017e-05, + "loss": 0.02, + "step": 2450 + }, + { + "epoch": 4.205128205128205, + "grad_norm": 0.09629535675048828, + "learning_rate": 2.8074290409161084e-05, + "loss": 0.0195, + "step": 2460 + }, + { + "epoch": 4.222222222222222, + "grad_norm": 0.0992991030216217, + "learning_rate": 2.8053978206289984e-05, + "loss": 0.0191, + "step": 2470 + }, + { + "epoch": 4.239316239316239, + "grad_norm": 0.0984906479716301, + "learning_rate": 2.803356686713139e-05, + "loss": 0.0191, + "step": 2480 + }, + { + "epoch": 4.256410256410256, + "grad_norm": 0.10978572815656662, + "learning_rate": 2.801305654669587e-05, + "loss": 0.0223, + "step": 2490 + }, + { + "epoch": 4.273504273504273, + "grad_norm": 0.09285573661327362, + "learning_rate": 2.799244740074566e-05, + "loss": 0.0195, + "step": 2500 + }, + { + "epoch": 4.2905982905982905, + "grad_norm": 0.08738566935062408, + "learning_rate": 2.797173958579353e-05, + "loss": 0.0228, + "step": 2510 + }, + { + "epoch": 4.3076923076923075, + "grad_norm": 0.11125430464744568, + "learning_rate": 2.7950933259101575e-05, + "loss": 0.0203, + "step": 2520 + }, + { + "epoch": 4.3247863247863245, + "grad_norm": 0.10201579332351685, + "learning_rate": 2.793002857868001e-05, + "loss": 0.0214, + "step": 2530 + }, + { + "epoch": 4.3418803418803416, + "grad_norm": 0.10354452580213547, + "learning_rate": 2.7909025703285983e-05, + "loss": 0.0199, + "step": 2540 + }, + { + "epoch": 4.358974358974359, + "grad_norm": 0.09892477840185165, + "learning_rate": 2.7887924792422385e-05, + "loss": 0.0224, + "step": 2550 + }, + { + "epoch": 4.3760683760683765, + "grad_norm": 0.1194918155670166, + "learning_rate": 2.78667260063366e-05, + "loss": 0.0231, + "step": 2560 + }, + { + "epoch": 4.3931623931623935, + "grad_norm": 0.11821082979440689, + "learning_rate": 2.784542950601931e-05, + "loss": 0.0196, + "step": 2570 + }, + { + "epoch": 4.410256410256411, + "grad_norm": 0.09250211715698242, + "learning_rate": 2.7824035453203293e-05, + "loss": 0.0203, + "step": 2580 + }, + { + "epoch": 4.427350427350428, + "grad_norm": 0.12260550260543823, + "learning_rate": 2.7802544010362154e-05, + "loss": 0.0186, + "step": 2590 + }, + { + "epoch": 4.444444444444445, + "grad_norm": 0.0743408128619194, + "learning_rate": 2.778095534070912e-05, + "loss": 0.0209, + "step": 2600 + }, + { + "epoch": 4.461538461538462, + "grad_norm": 0.08710762113332748, + "learning_rate": 2.7759269608195777e-05, + "loss": 0.0204, + "step": 2610 + }, + { + "epoch": 4.478632478632479, + "grad_norm": 0.08449119329452515, + "learning_rate": 2.773748697751086e-05, + "loss": 0.0199, + "step": 2620 + }, + { + "epoch": 4.495726495726496, + "grad_norm": 0.09476049244403839, + "learning_rate": 2.7715607614078956e-05, + "loss": 0.0201, + "step": 2630 + }, + { + "epoch": 4.512820512820513, + "grad_norm": 0.09813626855611801, + "learning_rate": 2.7693631684059305e-05, + "loss": 0.0204, + "step": 2640 + }, + { + "epoch": 4.52991452991453, + "grad_norm": 0.10010302066802979, + "learning_rate": 2.767155935434448e-05, + "loss": 0.0194, + "step": 2650 + }, + { + "epoch": 4.547008547008547, + "grad_norm": 0.1183779239654541, + "learning_rate": 2.7649390792559146e-05, + "loss": 0.0203, + "step": 2660 + }, + { + "epoch": 4.564102564102564, + "grad_norm": 0.08748367428779602, + "learning_rate": 2.7627126167058807e-05, + "loss": 0.0202, + "step": 2670 + }, + { + "epoch": 4.581196581196581, + "grad_norm": 0.12158568203449249, + "learning_rate": 2.7604765646928482e-05, + "loss": 0.0207, + "step": 2680 + }, + { + "epoch": 4.598290598290598, + "grad_norm": 0.1865331381559372, + "learning_rate": 2.758230940198147e-05, + "loss": 0.0208, + "step": 2690 + }, + { + "epoch": 4.615384615384615, + "grad_norm": 0.1120152547955513, + "learning_rate": 2.755975760275802e-05, + "loss": 0.0199, + "step": 2700 + }, + { + "epoch": 4.632478632478632, + "grad_norm": 0.08776689320802689, + "learning_rate": 2.7537110420524057e-05, + "loss": 0.0193, + "step": 2710 + }, + { + "epoch": 4.64957264957265, + "grad_norm": 0.13004325330257416, + "learning_rate": 2.7514368027269875e-05, + "loss": 0.0187, + "step": 2720 + }, + { + "epoch": 4.666666666666667, + "grad_norm": 0.09902483224868774, + "learning_rate": 2.749153059570883e-05, + "loss": 0.0201, + "step": 2730 + }, + { + "epoch": 4.683760683760684, + "grad_norm": 0.1488153040409088, + "learning_rate": 2.7468598299276044e-05, + "loss": 0.0191, + "step": 2740 + }, + { + "epoch": 4.700854700854701, + "grad_norm": 0.10970397293567657, + "learning_rate": 2.7445571312127058e-05, + "loss": 0.0184, + "step": 2750 + }, + { + "epoch": 4.717948717948718, + "grad_norm": 0.09405538439750671, + "learning_rate": 2.7422449809136538e-05, + "loss": 0.0195, + "step": 2760 + }, + { + "epoch": 4.735042735042735, + "grad_norm": 0.09082396328449249, + "learning_rate": 2.739923396589693e-05, + "loss": 0.0204, + "step": 2770 + }, + { + "epoch": 4.752136752136752, + "grad_norm": 0.1001177504658699, + "learning_rate": 2.737592395871713e-05, + "loss": 0.0173, + "step": 2780 + }, + { + "epoch": 4.769230769230769, + "grad_norm": 0.09204474091529846, + "learning_rate": 2.7352519964621146e-05, + "loss": 0.0185, + "step": 2790 + }, + { + "epoch": 4.786324786324786, + "grad_norm": 0.09500245749950409, + "learning_rate": 2.7329022161346758e-05, + "loss": 0.0198, + "step": 2800 + }, + { + "epoch": 4.803418803418803, + "grad_norm": 0.08855962008237839, + "learning_rate": 2.730543072734417e-05, + "loss": 0.0209, + "step": 2810 + }, + { + "epoch": 4.82051282051282, + "grad_norm": 0.09660346806049347, + "learning_rate": 2.728174584177463e-05, + "loss": 0.0197, + "step": 2820 + }, + { + "epoch": 4.837606837606837, + "grad_norm": 0.11314470320940018, + "learning_rate": 2.725796768450911e-05, + "loss": 0.0181, + "step": 2830 + }, + { + "epoch": 4.854700854700854, + "grad_norm": 0.09389517456293106, + "learning_rate": 2.7234096436126903e-05, + "loss": 0.02, + "step": 2840 + }, + { + "epoch": 4.871794871794872, + "grad_norm": 0.09539874643087387, + "learning_rate": 2.721013227791428e-05, + "loss": 0.0209, + "step": 2850 + }, + { + "epoch": 4.888888888888889, + "grad_norm": 0.09690243750810623, + "learning_rate": 2.7186075391863088e-05, + "loss": 0.0194, + "step": 2860 + }, + { + "epoch": 4.905982905982906, + "grad_norm": 0.10651098191738129, + "learning_rate": 2.7161925960669383e-05, + "loss": 0.0219, + "step": 2870 + }, + { + "epoch": 4.923076923076923, + "grad_norm": 0.12198858708143234, + "learning_rate": 2.713768416773205e-05, + "loss": 0.0194, + "step": 2880 + }, + { + "epoch": 4.94017094017094, + "grad_norm": 0.0976480171084404, + "learning_rate": 2.7113350197151394e-05, + "loss": 0.0203, + "step": 2890 + }, + { + "epoch": 4.957264957264957, + "grad_norm": 0.10814037919044495, + "learning_rate": 2.708892423372774e-05, + "loss": 0.0208, + "step": 2900 + }, + { + "epoch": 4.9743589743589745, + "grad_norm": 0.08893640339374542, + "learning_rate": 2.7064406462960044e-05, + "loss": 0.0167, + "step": 2910 + }, + { + "epoch": 4.9914529914529915, + "grad_norm": 0.1479610800743103, + "learning_rate": 2.7039797071044482e-05, + "loss": 0.022, + "step": 2920 + }, + { + "epoch": 5.0085470085470085, + "grad_norm": 0.09371191263198853, + "learning_rate": 2.7015096244873028e-05, + "loss": 0.0199, + "step": 2930 + }, + { + "epoch": 5.0256410256410255, + "grad_norm": 0.12326940894126892, + "learning_rate": 2.6990304172032038e-05, + "loss": 0.0195, + "step": 2940 + }, + { + "epoch": 5.042735042735043, + "grad_norm": 0.1065201461315155, + "learning_rate": 2.6965421040800823e-05, + "loss": 0.02, + "step": 2950 + }, + { + "epoch": 5.05982905982906, + "grad_norm": 0.059518881142139435, + "learning_rate": 2.6940447040150228e-05, + "loss": 0.0183, + "step": 2960 + }, + { + "epoch": 5.076923076923077, + "grad_norm": 0.1376444697380066, + "learning_rate": 2.691538235974118e-05, + "loss": 0.0196, + "step": 2970 + }, + { + "epoch": 5.094017094017094, + "grad_norm": 0.09283936023712158, + "learning_rate": 2.6890227189923285e-05, + "loss": 0.0176, + "step": 2980 + }, + { + "epoch": 5.111111111111111, + "grad_norm": 0.10842864215373993, + "learning_rate": 2.6864981721733317e-05, + "loss": 0.0177, + "step": 2990 + }, + { + "epoch": 5.128205128205128, + "grad_norm": 0.10565061122179031, + "learning_rate": 2.683964614689383e-05, + "loss": 0.0188, + "step": 3000 + }, + { + "epoch": 5.145299145299146, + "grad_norm": 0.09310047328472137, + "learning_rate": 2.681422065781169e-05, + "loss": 0.0171, + "step": 3010 + }, + { + "epoch": 5.162393162393163, + "grad_norm": 0.10931243002414703, + "learning_rate": 2.6788705447576563e-05, + "loss": 0.0186, + "step": 3020 + }, + { + "epoch": 5.17948717948718, + "grad_norm": 0.082731693983078, + "learning_rate": 2.6763100709959524e-05, + "loss": 0.0164, + "step": 3030 + }, + { + "epoch": 5.196581196581197, + "grad_norm": 0.13504242897033691, + "learning_rate": 2.6737406639411537e-05, + "loss": 0.0194, + "step": 3040 + }, + { + "epoch": 5.213675213675214, + "grad_norm": 0.0828668624162674, + "learning_rate": 2.6711623431061985e-05, + "loss": 0.0159, + "step": 3050 + }, + { + "epoch": 5.230769230769231, + "grad_norm": 0.10685226321220398, + "learning_rate": 2.6685751280717195e-05, + "loss": 0.0173, + "step": 3060 + }, + { + "epoch": 5.247863247863248, + "grad_norm": 0.09575466811656952, + "learning_rate": 2.665979038485896e-05, + "loss": 0.0182, + "step": 3070 + }, + { + "epoch": 5.264957264957265, + "grad_norm": 0.09520144015550613, + "learning_rate": 2.6633740940643022e-05, + "loss": 0.0183, + "step": 3080 + }, + { + "epoch": 5.282051282051282, + "grad_norm": 0.07664638757705688, + "learning_rate": 2.6607603145897594e-05, + "loss": 0.019, + "step": 3090 + }, + { + "epoch": 5.299145299145299, + "grad_norm": 0.10608942806720734, + "learning_rate": 2.6581377199121853e-05, + "loss": 0.0221, + "step": 3100 + }, + { + "epoch": 5.316239316239316, + "grad_norm": 0.08229487389326096, + "learning_rate": 2.6555063299484434e-05, + "loss": 0.0176, + "step": 3110 + }, + { + "epoch": 5.333333333333333, + "grad_norm": 0.0870119258761406, + "learning_rate": 2.6528661646821917e-05, + "loss": 0.0177, + "step": 3120 + }, + { + "epoch": 5.35042735042735, + "grad_norm": 0.13353557884693146, + "learning_rate": 2.6502172441637302e-05, + "loss": 0.0188, + "step": 3130 + }, + { + "epoch": 5.367521367521368, + "grad_norm": 0.09941720962524414, + "learning_rate": 2.6475595885098493e-05, + "loss": 0.0183, + "step": 3140 + }, + { + "epoch": 5.384615384615385, + "grad_norm": 0.09147895872592926, + "learning_rate": 2.644893217903677e-05, + "loss": 0.0195, + "step": 3150 + }, + { + "epoch": 5.401709401709402, + "grad_norm": 0.08084908127784729, + "learning_rate": 2.6422181525945265e-05, + "loss": 0.0164, + "step": 3160 + }, + { + "epoch": 5.418803418803419, + "grad_norm": 0.09985609352588654, + "learning_rate": 2.63953441289774e-05, + "loss": 0.0185, + "step": 3170 + }, + { + "epoch": 5.435897435897436, + "grad_norm": 0.10765116661787033, + "learning_rate": 2.6368420191945365e-05, + "loss": 0.0175, + "step": 3180 + }, + { + "epoch": 5.452991452991453, + "grad_norm": 0.07902234047651291, + "learning_rate": 2.6341409919318572e-05, + "loss": 0.0176, + "step": 3190 + }, + { + "epoch": 5.47008547008547, + "grad_norm": 0.08553142845630646, + "learning_rate": 2.6314313516222084e-05, + "loss": 0.0197, + "step": 3200 + }, + { + "epoch": 5.487179487179487, + "grad_norm": 0.10211332142353058, + "learning_rate": 2.6287131188435075e-05, + "loss": 0.0192, + "step": 3210 + }, + { + "epoch": 5.504273504273504, + "grad_norm": 0.09006617218255997, + "learning_rate": 2.6259863142389252e-05, + "loss": 0.0164, + "step": 3220 + }, + { + "epoch": 5.521367521367521, + "grad_norm": 0.09639620035886765, + "learning_rate": 2.62325095851673e-05, + "loss": 0.0151, + "step": 3230 + }, + { + "epoch": 5.538461538461538, + "grad_norm": 0.07281971722841263, + "learning_rate": 2.620507072450131e-05, + "loss": 0.0183, + "step": 3240 + }, + { + "epoch": 5.555555555555555, + "grad_norm": 0.06957155466079712, + "learning_rate": 2.6177546768771183e-05, + "loss": 0.0172, + "step": 3250 + }, + { + "epoch": 5.572649572649572, + "grad_norm": 0.08119498938322067, + "learning_rate": 2.6149937927003062e-05, + "loss": 0.019, + "step": 3260 + }, + { + "epoch": 5.589743589743589, + "grad_norm": 0.07163924723863602, + "learning_rate": 2.6122244408867766e-05, + "loss": 0.0194, + "step": 3270 + }, + { + "epoch": 5.6068376068376065, + "grad_norm": 0.09349033981561661, + "learning_rate": 2.6094466424679148e-05, + "loss": 0.0189, + "step": 3280 + }, + { + "epoch": 5.6239316239316235, + "grad_norm": 0.08922473341226578, + "learning_rate": 2.606660418539253e-05, + "loss": 0.0215, + "step": 3290 + }, + { + "epoch": 5.641025641025641, + "grad_norm": 0.10527209937572479, + "learning_rate": 2.6038657902603107e-05, + "loss": 0.018, + "step": 3300 + }, + { + "epoch": 5.6581196581196584, + "grad_norm": 0.1067451685667038, + "learning_rate": 2.6010627788544314e-05, + "loss": 0.0191, + "step": 3310 + }, + { + "epoch": 5.6752136752136755, + "grad_norm": 0.10629488527774811, + "learning_rate": 2.5982514056086238e-05, + "loss": 0.0182, + "step": 3320 + }, + { + "epoch": 5.6923076923076925, + "grad_norm": 0.11531593650579453, + "learning_rate": 2.5954316918733992e-05, + "loss": 0.0187, + "step": 3330 + }, + { + "epoch": 5.7094017094017095, + "grad_norm": 0.11777219921350479, + "learning_rate": 2.5926036590626093e-05, + "loss": 0.0192, + "step": 3340 + }, + { + "epoch": 5.726495726495727, + "grad_norm": 0.11960875242948532, + "learning_rate": 2.589767328653283e-05, + "loss": 0.0183, + "step": 3350 + }, + { + "epoch": 5.743589743589744, + "grad_norm": 0.08489462733268738, + "learning_rate": 2.5869227221854646e-05, + "loss": 0.0179, + "step": 3360 + }, + { + "epoch": 5.760683760683761, + "grad_norm": 0.07769697159528732, + "learning_rate": 2.584069861262049e-05, + "loss": 0.0186, + "step": 3370 + }, + { + "epoch": 5.777777777777778, + "grad_norm": 0.10025164484977722, + "learning_rate": 2.58120876754862e-05, + "loss": 0.0143, + "step": 3380 + }, + { + "epoch": 5.794871794871795, + "grad_norm": 0.09745113551616669, + "learning_rate": 2.5783394627732815e-05, + "loss": 0.0164, + "step": 3390 + }, + { + "epoch": 5.811965811965812, + "grad_norm": 0.09607358276844025, + "learning_rate": 2.5754619687264957e-05, + "loss": 0.0205, + "step": 3400 + }, + { + "epoch": 5.829059829059829, + "grad_norm": 0.0831088125705719, + "learning_rate": 2.5725763072609178e-05, + "loss": 0.0166, + "step": 3410 + }, + { + "epoch": 5.846153846153846, + "grad_norm": 0.07195224612951279, + "learning_rate": 2.569682500291229e-05, + "loss": 0.0152, + "step": 3420 + }, + { + "epoch": 5.863247863247864, + "grad_norm": 0.09906742721796036, + "learning_rate": 2.5667805697939687e-05, + "loss": 0.0162, + "step": 3430 + }, + { + "epoch": 5.880341880341881, + "grad_norm": 0.07969392836093903, + "learning_rate": 2.5638705378073704e-05, + "loss": 0.0193, + "step": 3440 + }, + { + "epoch": 5.897435897435898, + "grad_norm": 0.0873081237077713, + "learning_rate": 2.5609524264311934e-05, + "loss": 0.0177, + "step": 3450 + }, + { + "epoch": 5.914529914529915, + "grad_norm": 0.0908493623137474, + "learning_rate": 2.5580262578265536e-05, + "loss": 0.0205, + "step": 3460 + }, + { + "epoch": 5.931623931623932, + "grad_norm": 0.08355225622653961, + "learning_rate": 2.5550920542157557e-05, + "loss": 0.0161, + "step": 3470 + }, + { + "epoch": 5.948717948717949, + "grad_norm": 0.06161191314458847, + "learning_rate": 2.5521498378821273e-05, + "loss": 0.0183, + "step": 3480 + }, + { + "epoch": 5.965811965811966, + "grad_norm": 0.09668168425559998, + "learning_rate": 2.549199631169844e-05, + "loss": 0.0175, + "step": 3490 + }, + { + "epoch": 5.982905982905983, + "grad_norm": 0.07232195138931274, + "learning_rate": 2.5462414564837663e-05, + "loss": 0.0186, + "step": 3500 + }, + { + "epoch": 6.0, + "grad_norm": 0.12467692792415619, + "learning_rate": 2.543275336289264e-05, + "loss": 0.017, + "step": 3510 + }, + { + "epoch": 6.017094017094017, + "grad_norm": 0.11001081019639969, + "learning_rate": 2.540301293112048e-05, + "loss": 0.0168, + "step": 3520 + }, + { + "epoch": 6.034188034188034, + "grad_norm": 0.08639950305223465, + "learning_rate": 2.537319349538e-05, + "loss": 0.0193, + "step": 3530 + }, + { + "epoch": 6.051282051282051, + "grad_norm": 0.09567685425281525, + "learning_rate": 2.534329528212999e-05, + "loss": 0.0168, + "step": 3540 + }, + { + "epoch": 6.068376068376068, + "grad_norm": 0.08502615988254547, + "learning_rate": 2.531331851842751e-05, + "loss": 0.0174, + "step": 3550 + }, + { + "epoch": 6.085470085470085, + "grad_norm": 0.07735824584960938, + "learning_rate": 2.528326343192615e-05, + "loss": 0.0172, + "step": 3560 + }, + { + "epoch": 6.102564102564102, + "grad_norm": 0.06508814543485641, + "learning_rate": 2.5253130250874316e-05, + "loss": 0.0169, + "step": 3570 + }, + { + "epoch": 6.119658119658119, + "grad_norm": 0.11148368567228317, + "learning_rate": 2.5222919204113494e-05, + "loss": 0.0188, + "step": 3580 + }, + { + "epoch": 6.136752136752137, + "grad_norm": 0.11212950944900513, + "learning_rate": 2.5192630521076494e-05, + "loss": 0.0176, + "step": 3590 + }, + { + "epoch": 6.153846153846154, + "grad_norm": 0.09221594035625458, + "learning_rate": 2.5162264431785734e-05, + "loss": 0.0201, + "step": 3600 + }, + { + "epoch": 6.170940170940171, + "grad_norm": 0.10451870411634445, + "learning_rate": 2.5131821166851484e-05, + "loss": 0.0165, + "step": 3610 + }, + { + "epoch": 6.188034188034188, + "grad_norm": 0.09675891697406769, + "learning_rate": 2.5101300957470095e-05, + "loss": 0.0165, + "step": 3620 + }, + { + "epoch": 6.205128205128205, + "grad_norm": 0.07100088149309158, + "learning_rate": 2.507070403542228e-05, + "loss": 0.0156, + "step": 3630 + }, + { + "epoch": 6.222222222222222, + "grad_norm": 0.09254378825426102, + "learning_rate": 2.504003063307132e-05, + "loss": 0.0171, + "step": 3640 + }, + { + "epoch": 6.239316239316239, + "grad_norm": 0.07285033911466599, + "learning_rate": 2.500928098336131e-05, + "loss": 0.017, + "step": 3650 + }, + { + "epoch": 6.256410256410256, + "grad_norm": 0.10864458978176117, + "learning_rate": 2.497845531981541e-05, + "loss": 0.0171, + "step": 3660 + }, + { + "epoch": 6.273504273504273, + "grad_norm": 0.09306647628545761, + "learning_rate": 2.4947553876534047e-05, + "loss": 0.0173, + "step": 3670 + }, + { + "epoch": 6.2905982905982905, + "grad_norm": 0.08375543355941772, + "learning_rate": 2.4916576888193137e-05, + "loss": 0.0158, + "step": 3680 + }, + { + "epoch": 6.3076923076923075, + "grad_norm": 0.10552448034286499, + "learning_rate": 2.488552459004231e-05, + "loss": 0.0174, + "step": 3690 + }, + { + "epoch": 6.3247863247863245, + "grad_norm": 0.07783381640911102, + "learning_rate": 2.4854397217903145e-05, + "loss": 0.0177, + "step": 3700 + }, + { + "epoch": 6.3418803418803416, + "grad_norm": 0.10520699620246887, + "learning_rate": 2.4823195008167325e-05, + "loss": 0.0173, + "step": 3710 + }, + { + "epoch": 6.358974358974359, + "grad_norm": 0.07988301664590836, + "learning_rate": 2.4791918197794907e-05, + "loss": 0.0173, + "step": 3720 + }, + { + "epoch": 6.3760683760683765, + "grad_norm": 0.09742307662963867, + "learning_rate": 2.4760567024312476e-05, + "loss": 0.0166, + "step": 3730 + }, + { + "epoch": 6.3931623931623935, + "grad_norm": 0.08026914298534393, + "learning_rate": 2.4729141725811344e-05, + "loss": 0.0183, + "step": 3740 + }, + { + "epoch": 6.410256410256411, + "grad_norm": 0.11827661842107773, + "learning_rate": 2.4697642540945774e-05, + "loss": 0.0182, + "step": 3750 + }, + { + "epoch": 6.427350427350428, + "grad_norm": 0.09172654151916504, + "learning_rate": 2.4666069708931136e-05, + "loss": 0.0175, + "step": 3760 + }, + { + "epoch": 6.444444444444445, + "grad_norm": 0.08934371173381805, + "learning_rate": 2.46344234695421e-05, + "loss": 0.0178, + "step": 3770 + }, + { + "epoch": 6.461538461538462, + "grad_norm": 0.0789095088839531, + "learning_rate": 2.460270406311082e-05, + "loss": 0.0165, + "step": 3780 + }, + { + "epoch": 6.478632478632479, + "grad_norm": 0.11394141614437103, + "learning_rate": 2.457091173052511e-05, + "loss": 0.0187, + "step": 3790 + }, + { + "epoch": 6.495726495726496, + "grad_norm": 0.08394981920719147, + "learning_rate": 2.4539046713226593e-05, + "loss": 0.0169, + "step": 3800 + }, + { + "epoch": 6.512820512820513, + "grad_norm": 0.08010982722043991, + "learning_rate": 2.4507109253208898e-05, + "loss": 0.0145, + "step": 3810 + }, + { + "epoch": 6.52991452991453, + "grad_norm": 0.09592076390981674, + "learning_rate": 2.4475099593015813e-05, + "loss": 0.0159, + "step": 3820 + }, + { + "epoch": 6.547008547008547, + "grad_norm": 0.0980319082736969, + "learning_rate": 2.4443017975739423e-05, + "loss": 0.018, + "step": 3830 + }, + { + "epoch": 6.564102564102564, + "grad_norm": 0.09010566025972366, + "learning_rate": 2.4410864645018288e-05, + "loss": 0.0159, + "step": 3840 + }, + { + "epoch": 6.581196581196581, + "grad_norm": 0.10003632307052612, + "learning_rate": 2.4378639845035583e-05, + "loss": 0.0168, + "step": 3850 + }, + { + "epoch": 6.598290598290598, + "grad_norm": 0.09281107783317566, + "learning_rate": 2.4346343820517237e-05, + "loss": 0.0158, + "step": 3860 + }, + { + "epoch": 6.615384615384615, + "grad_norm": 0.10115903615951538, + "learning_rate": 2.4313976816730095e-05, + "loss": 0.0175, + "step": 3870 + }, + { + "epoch": 6.632478632478632, + "grad_norm": 0.07335299998521805, + "learning_rate": 2.4281539079480047e-05, + "loss": 0.0145, + "step": 3880 + }, + { + "epoch": 6.64957264957265, + "grad_norm": 0.09117656201124191, + "learning_rate": 2.4249030855110128e-05, + "loss": 0.0157, + "step": 3890 + }, + { + "epoch": 6.666666666666667, + "grad_norm": 0.0800071731209755, + "learning_rate": 2.42164523904987e-05, + "loss": 0.0165, + "step": 3900 + }, + { + "epoch": 6.683760683760684, + "grad_norm": 0.0696878433227539, + "learning_rate": 2.4183803933057554e-05, + "loss": 0.0174, + "step": 3910 + }, + { + "epoch": 6.700854700854701, + "grad_norm": 0.0962003841996193, + "learning_rate": 2.4151085730730013e-05, + "loss": 0.0182, + "step": 3920 + }, + { + "epoch": 6.717948717948718, + "grad_norm": 0.0997065007686615, + "learning_rate": 2.4118298031989084e-05, + "loss": 0.0175, + "step": 3930 + }, + { + "epoch": 6.735042735042735, + "grad_norm": 0.10266530513763428, + "learning_rate": 2.4085441085835534e-05, + "loss": 0.0164, + "step": 3940 + }, + { + "epoch": 6.752136752136752, + "grad_norm": 0.09127513319253922, + "learning_rate": 2.405251514179604e-05, + "loss": 0.0175, + "step": 3950 + }, + { + "epoch": 6.769230769230769, + "grad_norm": 0.06895238906145096, + "learning_rate": 2.401952044992126e-05, + "loss": 0.017, + "step": 3960 + }, + { + "epoch": 6.786324786324786, + "grad_norm": 0.08264075219631195, + "learning_rate": 2.3986457260783942e-05, + "loss": 0.0175, + "step": 3970 + }, + { + "epoch": 6.803418803418803, + "grad_norm": 0.13420195877552032, + "learning_rate": 2.395332582547703e-05, + "loss": 0.0209, + "step": 3980 + }, + { + "epoch": 6.82051282051282, + "grad_norm": 0.10509937256574631, + "learning_rate": 2.3920126395611755e-05, + "loss": 0.0182, + "step": 3990 + }, + { + "epoch": 6.837606837606837, + "grad_norm": 0.08453608304262161, + "learning_rate": 2.3886859223315725e-05, + "loss": 0.0183, + "step": 4000 + }, + { + "epoch": 6.854700854700854, + "grad_norm": 0.07713332772254944, + "learning_rate": 2.385352456123099e-05, + "loss": 0.0168, + "step": 4010 + }, + { + "epoch": 6.871794871794872, + "grad_norm": 0.08435060828924179, + "learning_rate": 2.3820122662512155e-05, + "loss": 0.017, + "step": 4020 + }, + { + "epoch": 6.888888888888889, + "grad_norm": 0.07585082948207855, + "learning_rate": 2.3786653780824443e-05, + "loss": 0.0158, + "step": 4030 + }, + { + "epoch": 6.905982905982906, + "grad_norm": 0.09102822840213776, + "learning_rate": 2.3753118170341762e-05, + "loss": 0.0163, + "step": 4040 + }, + { + "epoch": 6.923076923076923, + "grad_norm": 0.08681073039770126, + "learning_rate": 2.3719516085744782e-05, + "loss": 0.0149, + "step": 4050 + }, + { + "epoch": 6.94017094017094, + "grad_norm": 0.09147541224956512, + "learning_rate": 2.3685847782218997e-05, + "loss": 0.0153, + "step": 4060 + }, + { + "epoch": 6.957264957264957, + "grad_norm": 0.11664493381977081, + "learning_rate": 2.3652113515452792e-05, + "loss": 0.0169, + "step": 4070 + }, + { + "epoch": 6.9743589743589745, + "grad_norm": 0.08450709283351898, + "learning_rate": 2.3618313541635506e-05, + "loss": 0.0172, + "step": 4080 + }, + { + "epoch": 6.9914529914529915, + "grad_norm": 0.09218313544988632, + "learning_rate": 2.358444811745547e-05, + "loss": 0.0145, + "step": 4090 + }, + { + "epoch": 7.0085470085470085, + "grad_norm": 0.1001223474740982, + "learning_rate": 2.3550517500098068e-05, + "loss": 0.0158, + "step": 4100 + }, + { + "epoch": 7.0256410256410255, + "grad_norm": 0.07710881531238556, + "learning_rate": 2.3516521947243787e-05, + "loss": 0.016, + "step": 4110 + }, + { + "epoch": 7.042735042735043, + "grad_norm": 0.0710505023598671, + "learning_rate": 2.348246171706626e-05, + "loss": 0.0151, + "step": 4120 + }, + { + "epoch": 7.05982905982906, + "grad_norm": 0.08820394426584244, + "learning_rate": 2.3448337068230288e-05, + "loss": 0.0154, + "step": 4130 + }, + { + "epoch": 7.076923076923077, + "grad_norm": 0.09810654073953629, + "learning_rate": 2.3414148259889896e-05, + "loss": 0.0179, + "step": 4140 + }, + { + "epoch": 7.094017094017094, + "grad_norm": 0.07964249700307846, + "learning_rate": 2.3379895551686358e-05, + "loss": 0.0166, + "step": 4150 + }, + { + "epoch": 7.111111111111111, + "grad_norm": 0.10602698475122452, + "learning_rate": 2.334557920374623e-05, + "loss": 0.0179, + "step": 4160 + }, + { + "epoch": 7.128205128205128, + "grad_norm": 0.06306713074445724, + "learning_rate": 2.3311199476679367e-05, + "loss": 0.0159, + "step": 4170 + }, + { + "epoch": 7.145299145299146, + "grad_norm": 0.09402914345264435, + "learning_rate": 2.327675663157693e-05, + "loss": 0.0164, + "step": 4180 + }, + { + "epoch": 7.162393162393163, + "grad_norm": 0.10625981539487839, + "learning_rate": 2.3242250930009446e-05, + "loss": 0.0157, + "step": 4190 + }, + { + "epoch": 7.17948717948718, + "grad_norm": 0.07685832679271698, + "learning_rate": 2.3207682634024777e-05, + "loss": 0.016, + "step": 4200 + }, + { + "epoch": 7.196581196581197, + "grad_norm": 0.07576439529657364, + "learning_rate": 2.3173052006146155e-05, + "loss": 0.0145, + "step": 4210 + }, + { + "epoch": 7.213675213675214, + "grad_norm": 0.0860842764377594, + "learning_rate": 2.313835930937018e-05, + "loss": 0.0152, + "step": 4220 + }, + { + "epoch": 7.230769230769231, + "grad_norm": 0.06694341450929642, + "learning_rate": 2.310360480716483e-05, + "loss": 0.0139, + "step": 4230 + }, + { + "epoch": 7.247863247863248, + "grad_norm": 0.07913471013307571, + "learning_rate": 2.3068788763467434e-05, + "loss": 0.0138, + "step": 4240 + }, + { + "epoch": 7.264957264957265, + "grad_norm": 0.0752611756324768, + "learning_rate": 2.3033911442682708e-05, + "loss": 0.0173, + "step": 4250 + }, + { + "epoch": 7.282051282051282, + "grad_norm": 0.09453887492418289, + "learning_rate": 2.299897310968072e-05, + "loss": 0.0161, + "step": 4260 + }, + { + "epoch": 7.299145299145299, + "grad_norm": 0.07289311289787292, + "learning_rate": 2.2963974029794893e-05, + "loss": 0.014, + "step": 4270 + }, + { + "epoch": 7.316239316239316, + "grad_norm": 0.07161838561296463, + "learning_rate": 2.2928914468819968e-05, + "loss": 0.0172, + "step": 4280 + }, + { + "epoch": 7.333333333333333, + "grad_norm": 0.09718061983585358, + "learning_rate": 2.2893794693010002e-05, + "loss": 0.0171, + "step": 4290 + }, + { + "epoch": 7.35042735042735, + "grad_norm": 0.08493733406066895, + "learning_rate": 2.2858614969076358e-05, + "loss": 0.0156, + "step": 4300 + }, + { + "epoch": 7.367521367521368, + "grad_norm": 0.09210729598999023, + "learning_rate": 2.2823375564185645e-05, + "loss": 0.0177, + "step": 4310 + }, + { + "epoch": 7.384615384615385, + "grad_norm": 0.08660814166069031, + "learning_rate": 2.278807674595773e-05, + "loss": 0.016, + "step": 4320 + }, + { + "epoch": 7.401709401709402, + "grad_norm": 0.0879073292016983, + "learning_rate": 2.275271878246367e-05, + "loss": 0.0155, + "step": 4330 + }, + { + "epoch": 7.418803418803419, + "grad_norm": 0.0978168323636055, + "learning_rate": 2.271730194222369e-05, + "loss": 0.0167, + "step": 4340 + }, + { + "epoch": 7.435897435897436, + "grad_norm": 0.10353145748376846, + "learning_rate": 2.2681826494205156e-05, + "loss": 0.0168, + "step": 4350 + }, + { + "epoch": 7.452991452991453, + "grad_norm": 0.08525838702917099, + "learning_rate": 2.2646292707820513e-05, + "loss": 0.0155, + "step": 4360 + }, + { + "epoch": 7.47008547008547, + "grad_norm": 0.09657790511846542, + "learning_rate": 2.2610700852925254e-05, + "loss": 0.0153, + "step": 4370 + }, + { + "epoch": 7.487179487179487, + "grad_norm": 0.10523585230112076, + "learning_rate": 2.2575051199815854e-05, + "loss": 0.0172, + "step": 4380 + }, + { + "epoch": 7.504273504273504, + "grad_norm": 0.07577922940254211, + "learning_rate": 2.253934401922773e-05, + "loss": 0.0167, + "step": 4390 + }, + { + "epoch": 7.521367521367521, + "grad_norm": 0.09458927065134048, + "learning_rate": 2.250357958233319e-05, + "loss": 0.0178, + "step": 4400 + }, + { + "epoch": 7.538461538461538, + "grad_norm": 0.08498671650886536, + "learning_rate": 2.2467758160739352e-05, + "loss": 0.0165, + "step": 4410 + }, + { + "epoch": 7.555555555555555, + "grad_norm": 0.07391520589590073, + "learning_rate": 2.2431880026486114e-05, + "loss": 0.0163, + "step": 4420 + }, + { + "epoch": 7.572649572649572, + "grad_norm": 0.08839573711156845, + "learning_rate": 2.2395945452044045e-05, + "loss": 0.0187, + "step": 4430 + }, + { + "epoch": 7.589743589743589, + "grad_norm": 0.09913501143455505, + "learning_rate": 2.2359954710312363e-05, + "loss": 0.0141, + "step": 4440 + }, + { + "epoch": 7.6068376068376065, + "grad_norm": 0.12251193076372147, + "learning_rate": 2.232390807461682e-05, + "loss": 0.016, + "step": 4450 + }, + { + "epoch": 7.6239316239316235, + "grad_norm": 0.10324403643608093, + "learning_rate": 2.2287805818707658e-05, + "loss": 0.0174, + "step": 4460 + }, + { + "epoch": 7.641025641025641, + "grad_norm": 0.08906015753746033, + "learning_rate": 2.2251648216757514e-05, + "loss": 0.0147, + "step": 4470 + }, + { + "epoch": 7.6581196581196584, + "grad_norm": 0.09674681723117828, + "learning_rate": 2.2215435543359337e-05, + "loss": 0.0149, + "step": 4480 + }, + { + "epoch": 7.6752136752136755, + "grad_norm": 0.0863393247127533, + "learning_rate": 2.2179168073524308e-05, + "loss": 0.0164, + "step": 4490 + }, + { + "epoch": 7.6923076923076925, + "grad_norm": 0.07127152383327484, + "learning_rate": 2.2142846082679763e-05, + "loss": 0.0164, + "step": 4500 + }, + { + "epoch": 7.7094017094017095, + "grad_norm": 0.07405145466327667, + "learning_rate": 2.2106469846667078e-05, + "loss": 0.0147, + "step": 4510 + }, + { + "epoch": 7.726495726495727, + "grad_norm": 0.0741042047739029, + "learning_rate": 2.2070039641739578e-05, + "loss": 0.0168, + "step": 4520 + }, + { + "epoch": 7.743589743589744, + "grad_norm": 0.09215082973241806, + "learning_rate": 2.2033555744560462e-05, + "loss": 0.0156, + "step": 4530 + }, + { + "epoch": 7.760683760683761, + "grad_norm": 0.06406484544277191, + "learning_rate": 2.1997018432200674e-05, + "loss": 0.0153, + "step": 4540 + }, + { + "epoch": 7.777777777777778, + "grad_norm": 0.0733003243803978, + "learning_rate": 2.1960427982136818e-05, + "loss": 0.0143, + "step": 4550 + }, + { + "epoch": 7.794871794871795, + "grad_norm": 0.07966963946819305, + "learning_rate": 2.1923784672249038e-05, + "loss": 0.0168, + "step": 4560 + }, + { + "epoch": 7.811965811965812, + "grad_norm": 0.08074156194925308, + "learning_rate": 2.1887088780818918e-05, + "loss": 0.014, + "step": 4570 + }, + { + "epoch": 7.829059829059829, + "grad_norm": 0.10418794304132462, + "learning_rate": 2.1850340586527354e-05, + "loss": 0.0171, + "step": 4580 + }, + { + "epoch": 7.846153846153846, + "grad_norm": 0.0782044306397438, + "learning_rate": 2.1813540368452468e-05, + "loss": 0.0151, + "step": 4590 + }, + { + "epoch": 7.863247863247864, + "grad_norm": 0.08316248655319214, + "learning_rate": 2.1776688406067447e-05, + "loss": 0.0157, + "step": 4600 + }, + { + "epoch": 7.880341880341881, + "grad_norm": 0.08307249844074249, + "learning_rate": 2.1739784979238447e-05, + "loss": 0.0154, + "step": 4610 + }, + { + "epoch": 7.897435897435898, + "grad_norm": 0.05799156054854393, + "learning_rate": 2.1702830368222478e-05, + "loss": 0.0165, + "step": 4620 + }, + { + "epoch": 7.914529914529915, + "grad_norm": 0.0774935930967331, + "learning_rate": 2.166582485366523e-05, + "loss": 0.0156, + "step": 4630 + }, + { + "epoch": 7.931623931623932, + "grad_norm": 0.07723759859800339, + "learning_rate": 2.1628768716599e-05, + "loss": 0.0147, + "step": 4640 + }, + { + "epoch": 7.948717948717949, + "grad_norm": 0.09672164171934128, + "learning_rate": 2.1591662238440515e-05, + "loss": 0.0155, + "step": 4650 + }, + { + "epoch": 7.965811965811966, + "grad_norm": 0.0799407809972763, + "learning_rate": 2.1554505700988813e-05, + "loss": 0.0139, + "step": 4660 + }, + { + "epoch": 7.982905982905983, + "grad_norm": 0.0932922214269638, + "learning_rate": 2.1517299386423094e-05, + "loss": 0.0158, + "step": 4670 + }, + { + "epoch": 8.0, + "grad_norm": 0.0904204398393631, + "learning_rate": 2.148004357730059e-05, + "loss": 0.0155, + "step": 4680 + }, + { + "epoch": 8.017094017094017, + "grad_norm": 0.08726400136947632, + "learning_rate": 2.1442738556554404e-05, + "loss": 0.0165, + "step": 4690 + }, + { + "epoch": 8.034188034188034, + "grad_norm": 0.08845978230237961, + "learning_rate": 2.1405384607491375e-05, + "loss": 0.0175, + "step": 4700 + }, + { + "epoch": 8.051282051282051, + "grad_norm": 0.08011137694120407, + "learning_rate": 2.1367982013789913e-05, + "loss": 0.0155, + "step": 4710 + }, + { + "epoch": 8.068376068376068, + "grad_norm": 0.11915645003318787, + "learning_rate": 2.133053105949786e-05, + "loss": 0.0154, + "step": 4720 + }, + { + "epoch": 8.085470085470085, + "grad_norm": 0.06125086173415184, + "learning_rate": 2.129303202903032e-05, + "loss": 0.0162, + "step": 4730 + }, + { + "epoch": 8.102564102564102, + "grad_norm": 0.07610971480607986, + "learning_rate": 2.1255485207167504e-05, + "loss": 0.0156, + "step": 4740 + }, + { + "epoch": 8.11965811965812, + "grad_norm": 0.11904177814722061, + "learning_rate": 2.1217890879052556e-05, + "loss": 0.0154, + "step": 4750 + }, + { + "epoch": 8.136752136752136, + "grad_norm": 0.06247622147202492, + "learning_rate": 2.1180249330189423e-05, + "loss": 0.0167, + "step": 4760 + }, + { + "epoch": 8.153846153846153, + "grad_norm": 0.0804535299539566, + "learning_rate": 2.1142560846440645e-05, + "loss": 0.0171, + "step": 4770 + }, + { + "epoch": 8.17094017094017, + "grad_norm": 0.08552727848291397, + "learning_rate": 2.1104825714025203e-05, + "loss": 0.016, + "step": 4780 + }, + { + "epoch": 8.188034188034187, + "grad_norm": 0.06268288195133209, + "learning_rate": 2.106704421951634e-05, + "loss": 0.0165, + "step": 4790 + }, + { + "epoch": 8.205128205128204, + "grad_norm": 0.0662277489900589, + "learning_rate": 2.1029216649839404e-05, + "loss": 0.0138, + "step": 4800 + }, + { + "epoch": 8.222222222222221, + "grad_norm": 0.08757280558347702, + "learning_rate": 2.099134329226963e-05, + "loss": 0.0162, + "step": 4810 + }, + { + "epoch": 8.239316239316238, + "grad_norm": 0.08624053746461868, + "learning_rate": 2.0953424434430006e-05, + "loss": 0.0154, + "step": 4820 + }, + { + "epoch": 8.256410256410255, + "grad_norm": 0.10608145594596863, + "learning_rate": 2.091546036428905e-05, + "loss": 0.0151, + "step": 4830 + }, + { + "epoch": 8.273504273504274, + "grad_norm": 0.06986302137374878, + "learning_rate": 2.0877451370158633e-05, + "loss": 0.0157, + "step": 4840 + }, + { + "epoch": 8.290598290598291, + "grad_norm": 0.08707921206951141, + "learning_rate": 2.083939774069181e-05, + "loss": 0.0141, + "step": 4850 + }, + { + "epoch": 8.307692307692308, + "grad_norm": 0.07429157197475433, + "learning_rate": 2.080129976488059e-05, + "loss": 0.0145, + "step": 4860 + }, + { + "epoch": 8.324786324786325, + "grad_norm": 0.0858520120382309, + "learning_rate": 2.0763157732053795e-05, + "loss": 0.0155, + "step": 4870 + }, + { + "epoch": 8.341880341880342, + "grad_norm": 0.096824049949646, + "learning_rate": 2.0724971931874796e-05, + "loss": 0.0148, + "step": 4880 + }, + { + "epoch": 8.35897435897436, + "grad_norm": 0.061589065939188004, + "learning_rate": 2.0686742654339363e-05, + "loss": 0.0148, + "step": 4890 + }, + { + "epoch": 8.376068376068377, + "grad_norm": 0.08161415159702301, + "learning_rate": 2.0648470189773455e-05, + "loss": 0.0142, + "step": 4900 + }, + { + "epoch": 8.393162393162394, + "grad_norm": 0.057990241795778275, + "learning_rate": 2.0610154828830997e-05, + "loss": 0.0139, + "step": 4910 + }, + { + "epoch": 8.41025641025641, + "grad_norm": 0.09365546703338623, + "learning_rate": 2.0571796862491686e-05, + "loss": 0.0167, + "step": 4920 + }, + { + "epoch": 8.427350427350428, + "grad_norm": 0.11589137464761734, + "learning_rate": 2.0533396582058783e-05, + "loss": 0.0169, + "step": 4930 + }, + { + "epoch": 8.444444444444445, + "grad_norm": 0.07843475043773651, + "learning_rate": 2.0494954279156893e-05, + "loss": 0.0138, + "step": 4940 + }, + { + "epoch": 8.461538461538462, + "grad_norm": 0.07571505755186081, + "learning_rate": 2.0456470245729748e-05, + "loss": 0.016, + "step": 4950 + }, + { + "epoch": 8.478632478632479, + "grad_norm": 0.09430881589651108, + "learning_rate": 2.0417944774038013e-05, + "loss": 0.0165, + "step": 4960 + }, + { + "epoch": 8.495726495726496, + "grad_norm": 0.08276867121458054, + "learning_rate": 2.0379378156657026e-05, + "loss": 0.0146, + "step": 4970 + }, + { + "epoch": 8.512820512820513, + "grad_norm": 0.07636648416519165, + "learning_rate": 2.0340770686474618e-05, + "loss": 0.0149, + "step": 4980 + }, + { + "epoch": 8.52991452991453, + "grad_norm": 0.08420944213867188, + "learning_rate": 2.030212265668886e-05, + "loss": 0.0137, + "step": 4990 + }, + { + "epoch": 8.547008547008547, + "grad_norm": 0.0849870890378952, + "learning_rate": 2.0263434360805847e-05, + "loss": 0.0169, + "step": 5000 + } + ], + "logging_steps": 10, + "max_steps": 12000, + "num_input_tokens_seen": 0, + "num_train_epochs": 21, + "save_steps": 12001, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-8000/config.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-8000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..47d7b77e4255263f3803a4b6f94ca6831e459b70 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-8000/config.json @@ -0,0 +1,64 @@ +{ + "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" + ], + "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", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-8000/experiment_cfg/metadata.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-8000/experiment_cfg/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..23dcfa871a88e56280f945357aef0296bd071711 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-8000/experiment_cfg/metadata.json @@ -0,0 +1,431 @@ +{ + "new_embodiment": { + "statistics": { + "state": { + "base_position": { + "max": [ + 7.778976917266846, + 0.33403685688972473, + 0.7223060727119446 + ], + "min": [ + -4.821934223175049, + -6.785110950469971, + 0.6994727253913879 + ], + "mean": [ + 2.577927505176127, + -2.1525618086425604, + 0.7003541659101092 + ], + "std": [ + 1.6720365948940972, + 1.3768414767482617, + 0.0009102921174316669 + ], + "q01": [ + -0.7605057497562686, + -5.518671947805291, + 0.6999764703640262 + ], + "q99": [ + 6.038879433964306, + -0.5168299981147954, + 0.7029182461269848 + ] + }, + "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, + 0.24782371371751144, + 0.5971093380149053 + ], + "std": [ + 0.0, + 0.0, + 0.6642559499983499, + 0.3751891785184018 + ], + "q01": [ + 0.0, + 0.0, + -1.0, + 4.194193576416142e-06 + ], + "q99": [ + 0.0, + 0.0, + 1.0, + 1.0 + ] + }, + "end_effector_position_relative": { + "max": [ + 0.8848381638526917, + 0.8569459915161133, + 1.053884744644165 + ], + "min": [ + -0.5828549861907959, + -0.8486236333847046, + -0.3446122705936432 + ], + "mean": [ + 0.25006971579930243, + -0.005716176032514759, + 0.4426027721228362 + ], + "std": [ + 0.15838384751435522, + 0.25141504139800513, + 0.20646816234620438 + ], + "q01": [ + -0.017003721077380903, + -0.47738320165276543, + 0.1917563409753314 + ], + "q99": [ + 0.5401029199137674, + 0.4706646609268599, + 0.6965586537529516 + ] + }, + "end_effector_rotation_relative": { + "max": [ + 0.9999949932098389, + 0.9997354745864868, + 0.9900747537612915, + 0.9735639095306396 + ], + "min": [ + -0.9999914169311523, + -0.9999884366989136, + -0.9975369572639465, + 6.754255821306288e-08 + ], + "mean": [ + -0.22531628862537245, + 0.045669797538445966, + -0.06098055586474368, + 0.12292005493282424 + ], + "std": [ + 0.8506341437448487, + 0.2880763358322842, + 0.30816573781739026, + 0.16366165154486362 + ], + "q01": [ + -0.9957510593732032, + -0.5584741016891311, + -0.5371647381159943, + 0.0013784815838340492 + ], + "q99": [ + 0.9899375209218226, + 0.6327878884078038, + 0.5071539024531994, + 0.43555975763914345 + ] + }, + "gripper_qpos": { + "max": [ + 0.05289580300450325, + 0.00937709677964449 + ], + "min": [ + -0.011669117026031017, + -0.06307756155729294 + ], + "mean": [ + 0.03196506153371015, + -0.03229791582989581 + ], + "std": [ + 0.01171640503534049, + 0.01140016755425199 + ], + "q01": [ + 0.006290779996034005, + -0.04059717954551848 + ], + "q99": [ + 0.04038165838477291, + -0.006363076424932449 + ] + } + }, + "action": { + "base_motion": { + "max": [ + 1.0, + 1.0, + 1.0, + 0.0 + ], + "min": [ + -1.0, + -1.0, + -1.0, + 0.0 + ], + "mean": [ + 0.006112990930475949, + 0.002504955933349057, + -0.001639409958610569, + 0.0 + ], + "std": [ + 0.11478077265783658, + 0.1002662368589144, + 0.0807258815139021, + 0.0 + ], + "q01": [ + -0.16969970921613756, + -0.19919537649977875, + -0.15012139866371593, + 0.0 + ], + "q99": [ + 0.2466625453958827, + 0.25040842772679267, + 0.12777701094549537, + 0.0 + ] + }, + "control_mode": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.9253067479599785 + ], + "std": [ + 0.3792397014797737 + ], + "q01": [ + -1.0 + ], + "q99": [ + -0.1974093068624371 + ] + }, + "end_effector_position": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + -0.006754222051360652, + 0.0027854853184013867, + -0.08087572130330477 + ], + "std": [ + 0.45779905420789546, + 0.40578975219522756, + 0.4062694098132682 + ], + "q01": [ + -0.968295960290036, + -0.9240402337368848, + -0.8249359802202183 + ], + "q99": [ + 0.8046634915851651, + 0.9212243025654695, + 0.838433711437593 + ] + }, + "end_effector_rotation": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.008633749251857347, + -0.020406457524911718, + -0.0027639446362883006 + ], + "std": [ + 0.11393972872407104, + 0.13300186268009506, + 0.12192288302920105 + ], + "q01": [ + -0.2668579353776187, + -0.40299251206771697, + -0.34712957892172586 + ], + "q99": [ + 0.32748281264383927, + 0.3179028969292779, + 0.33459405510046736 + ] + }, + "gripper_close": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.3091635769453286 + ], + "std": [ + 0.9510122432876397 + ], + "q01": [ + -1.0 + ], + "q99": [ + 0.7643545615216344 + ] + } + } + }, + "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/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-8000/model.safetensors.index.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-8000/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..5ef5f247b61eb93e89601c7da8334fc087e19750 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-8000/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/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-8000/trainer_state.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-8000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..7e75448de063a28c35472bf0c128b733972af5dc --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/checkpoint-8000/trainer_state.json @@ -0,0 +1,5634 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 13.675213675213675, + "eval_steps": 500, + "global_step": 8000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.017094017094017096, + "grad_norm": 1.0089545249938965, + "learning_rate": 4.5e-07, + "loss": 0.2453, + "step": 10 + }, + { + "epoch": 0.03418803418803419, + "grad_norm": 0.8558177947998047, + "learning_rate": 9.500000000000001e-07, + "loss": 0.2334, + "step": 20 + }, + { + "epoch": 0.05128205128205128, + "grad_norm": 0.7379946112632751, + "learning_rate": 1.45e-06, + "loss": 0.2502, + "step": 30 + }, + { + "epoch": 0.06837606837606838, + "grad_norm": 0.8863121867179871, + "learning_rate": 1.95e-06, + "loss": 0.2143, + "step": 40 + }, + { + "epoch": 0.08547008547008547, + "grad_norm": 1.415384292602539, + "learning_rate": 2.45e-06, + "loss": 0.2502, + "step": 50 + }, + { + "epoch": 0.10256410256410256, + "grad_norm": 0.9582884311676025, + "learning_rate": 2.9499999999999997e-06, + "loss": 0.2322, + "step": 60 + }, + { + "epoch": 0.11965811965811966, + "grad_norm": 0.8592327237129211, + "learning_rate": 3.4500000000000004e-06, + "loss": 0.2117, + "step": 70 + }, + { + "epoch": 0.13675213675213677, + "grad_norm": 0.8079739212989807, + "learning_rate": 3.9499999999999995e-06, + "loss": 0.2001, + "step": 80 + }, + { + "epoch": 0.15384615384615385, + "grad_norm": 0.746914267539978, + "learning_rate": 4.450000000000001e-06, + "loss": 0.2006, + "step": 90 + }, + { + "epoch": 0.17094017094017094, + "grad_norm": 1.141075849533081, + "learning_rate": 4.95e-06, + "loss": 0.1755, + "step": 100 + }, + { + "epoch": 0.18803418803418803, + "grad_norm": 0.9814507365226746, + "learning_rate": 5.45e-06, + "loss": 0.1677, + "step": 110 + }, + { + "epoch": 0.20512820512820512, + "grad_norm": 0.5535752773284912, + "learning_rate": 5.95e-06, + "loss": 0.1627, + "step": 120 + }, + { + "epoch": 0.2222222222222222, + "grad_norm": 0.7861893773078918, + "learning_rate": 6.45e-06, + "loss": 0.1382, + "step": 130 + }, + { + "epoch": 0.23931623931623933, + "grad_norm": 0.43815529346466064, + "learning_rate": 6.95e-06, + "loss": 0.1404, + "step": 140 + }, + { + "epoch": 0.2564102564102564, + "grad_norm": 0.6113188862800598, + "learning_rate": 7.45e-06, + "loss": 0.1303, + "step": 150 + }, + { + "epoch": 0.27350427350427353, + "grad_norm": 0.4023075997829437, + "learning_rate": 7.95e-06, + "loss": 0.1121, + "step": 160 + }, + { + "epoch": 0.2905982905982906, + "grad_norm": 0.43074819445610046, + "learning_rate": 8.45e-06, + "loss": 0.1078, + "step": 170 + }, + { + "epoch": 0.3076923076923077, + "grad_norm": 0.28252357244491577, + "learning_rate": 8.95e-06, + "loss": 0.0936, + "step": 180 + }, + { + "epoch": 0.3247863247863248, + "grad_norm": 0.23254995048046112, + "learning_rate": 9.450000000000001e-06, + "loss": 0.0877, + "step": 190 + }, + { + "epoch": 0.3418803418803419, + "grad_norm": 0.20975816249847412, + "learning_rate": 9.95e-06, + "loss": 0.083, + "step": 200 + }, + { + "epoch": 0.358974358974359, + "grad_norm": 0.27841970324516296, + "learning_rate": 1.045e-05, + "loss": 0.0863, + "step": 210 + }, + { + "epoch": 0.37606837606837606, + "grad_norm": 0.2387871891260147, + "learning_rate": 1.095e-05, + "loss": 0.0822, + "step": 220 + }, + { + "epoch": 0.39316239316239315, + "grad_norm": 0.34779876470565796, + "learning_rate": 1.145e-05, + "loss": 0.0802, + "step": 230 + }, + { + "epoch": 0.41025641025641024, + "grad_norm": 0.2469312846660614, + "learning_rate": 1.1949999999999999e-05, + "loss": 0.0801, + "step": 240 + }, + { + "epoch": 0.42735042735042733, + "grad_norm": 0.18633294105529785, + "learning_rate": 1.245e-05, + "loss": 0.0736, + "step": 250 + }, + { + "epoch": 0.4444444444444444, + "grad_norm": 0.18194709718227386, + "learning_rate": 1.295e-05, + "loss": 0.0722, + "step": 260 + }, + { + "epoch": 0.46153846153846156, + "grad_norm": 0.21349109709262848, + "learning_rate": 1.345e-05, + "loss": 0.0732, + "step": 270 + }, + { + "epoch": 0.47863247863247865, + "grad_norm": 0.17933063209056854, + "learning_rate": 1.395e-05, + "loss": 0.0734, + "step": 280 + }, + { + "epoch": 0.49572649572649574, + "grad_norm": 0.18863998353481293, + "learning_rate": 1.445e-05, + "loss": 0.0673, + "step": 290 + }, + { + "epoch": 0.5128205128205128, + "grad_norm": 0.22107668220996857, + "learning_rate": 1.4950000000000001e-05, + "loss": 0.0701, + "step": 300 + }, + { + "epoch": 0.5299145299145299, + "grad_norm": 0.18152156472206116, + "learning_rate": 1.545e-05, + "loss": 0.0665, + "step": 310 + }, + { + "epoch": 0.5470085470085471, + "grad_norm": 0.1960478127002716, + "learning_rate": 1.5949999999999998e-05, + "loss": 0.0636, + "step": 320 + }, + { + "epoch": 0.5641025641025641, + "grad_norm": 0.21717658638954163, + "learning_rate": 1.645e-05, + "loss": 0.0644, + "step": 330 + }, + { + "epoch": 0.5811965811965812, + "grad_norm": 0.21584761142730713, + "learning_rate": 1.695e-05, + "loss": 0.0707, + "step": 340 + }, + { + "epoch": 0.5982905982905983, + "grad_norm": 0.22434745728969574, + "learning_rate": 1.745e-05, + "loss": 0.0651, + "step": 350 + }, + { + "epoch": 0.6153846153846154, + "grad_norm": 0.22187460958957672, + "learning_rate": 1.7950000000000003e-05, + "loss": 0.0636, + "step": 360 + }, + { + "epoch": 0.6324786324786325, + "grad_norm": 0.2252669632434845, + "learning_rate": 1.845e-05, + "loss": 0.0602, + "step": 370 + }, + { + "epoch": 0.6495726495726496, + "grad_norm": 0.1939157098531723, + "learning_rate": 1.8950000000000003e-05, + "loss": 0.0581, + "step": 380 + }, + { + "epoch": 0.6666666666666666, + "grad_norm": 0.19263336062431335, + "learning_rate": 1.945e-05, + "loss": 0.058, + "step": 390 + }, + { + "epoch": 0.6837606837606838, + "grad_norm": 0.1859099417924881, + "learning_rate": 1.995e-05, + "loss": 0.0646, + "step": 400 + }, + { + "epoch": 0.7008547008547008, + "grad_norm": 0.16536593437194824, + "learning_rate": 2.045e-05, + "loss": 0.0571, + "step": 410 + }, + { + "epoch": 0.717948717948718, + "grad_norm": 0.18150511384010315, + "learning_rate": 2.095e-05, + "loss": 0.0604, + "step": 420 + }, + { + "epoch": 0.7350427350427351, + "grad_norm": 0.19648237526416779, + "learning_rate": 2.145e-05, + "loss": 0.0607, + "step": 430 + }, + { + "epoch": 0.7521367521367521, + "grad_norm": 0.19251717627048492, + "learning_rate": 2.195e-05, + "loss": 0.0576, + "step": 440 + }, + { + "epoch": 0.7692307692307693, + "grad_norm": 0.173828586935997, + "learning_rate": 2.245e-05, + "loss": 0.0526, + "step": 450 + }, + { + "epoch": 0.7863247863247863, + "grad_norm": 0.22500741481781006, + "learning_rate": 2.2950000000000002e-05, + "loss": 0.057, + "step": 460 + }, + { + "epoch": 0.8034188034188035, + "grad_norm": 0.20360895991325378, + "learning_rate": 2.345e-05, + "loss": 0.0563, + "step": 470 + }, + { + "epoch": 0.8205128205128205, + "grad_norm": 0.1981644481420517, + "learning_rate": 2.395e-05, + "loss": 0.0498, + "step": 480 + }, + { + "epoch": 0.8376068376068376, + "grad_norm": 0.18252193927764893, + "learning_rate": 2.4449999999999998e-05, + "loss": 0.0532, + "step": 490 + }, + { + "epoch": 0.8547008547008547, + "grad_norm": 0.24334463477134705, + "learning_rate": 2.495e-05, + "loss": 0.054, + "step": 500 + }, + { + "epoch": 0.8717948717948718, + "grad_norm": 0.20569105446338654, + "learning_rate": 2.5450000000000002e-05, + "loss": 0.0474, + "step": 510 + }, + { + "epoch": 0.8888888888888888, + "grad_norm": 0.16904938220977783, + "learning_rate": 2.595e-05, + "loss": 0.0527, + "step": 520 + }, + { + "epoch": 0.905982905982906, + "grad_norm": 0.1478479504585266, + "learning_rate": 2.6450000000000003e-05, + "loss": 0.0494, + "step": 530 + }, + { + "epoch": 0.9230769230769231, + "grad_norm": 0.18176110088825226, + "learning_rate": 2.695e-05, + "loss": 0.0486, + "step": 540 + }, + { + "epoch": 0.9401709401709402, + "grad_norm": 0.19104203581809998, + "learning_rate": 2.7450000000000003e-05, + "loss": 0.0447, + "step": 550 + }, + { + "epoch": 0.9572649572649573, + "grad_norm": 0.21148279309272766, + "learning_rate": 2.795e-05, + "loss": 0.045, + "step": 560 + }, + { + "epoch": 0.9743589743589743, + "grad_norm": 0.1891864538192749, + "learning_rate": 2.845e-05, + "loss": 0.049, + "step": 570 + }, + { + "epoch": 0.9914529914529915, + "grad_norm": 0.18104152381420135, + "learning_rate": 2.895e-05, + "loss": 0.0461, + "step": 580 + }, + { + "epoch": 1.0085470085470085, + "grad_norm": 0.244981586933136, + "learning_rate": 2.945e-05, + "loss": 0.044, + "step": 590 + }, + { + "epoch": 1.0256410256410255, + "grad_norm": 0.1661752462387085, + "learning_rate": 2.995e-05, + "loss": 0.0464, + "step": 600 + }, + { + "epoch": 1.0427350427350428, + "grad_norm": 0.19394716620445251, + "learning_rate": 2.9999953864407506e-05, + "loss": 0.0434, + "step": 610 + }, + { + "epoch": 1.0598290598290598, + "grad_norm": 0.15391552448272705, + "learning_rate": 2.99997943837114e-05, + "loss": 0.0429, + "step": 620 + }, + { + "epoch": 1.0769230769230769, + "grad_norm": 0.17904387414455414, + "learning_rate": 2.9999520989547326e-05, + "loss": 0.0443, + "step": 630 + }, + { + "epoch": 1.0940170940170941, + "grad_norm": 0.1621219515800476, + "learning_rate": 2.9999133683991525e-05, + "loss": 0.0423, + "step": 640 + }, + { + "epoch": 1.1111111111111112, + "grad_norm": 0.1549968719482422, + "learning_rate": 2.9998632469985326e-05, + "loss": 0.0439, + "step": 650 + }, + { + "epoch": 1.1282051282051282, + "grad_norm": 0.17623211443424225, + "learning_rate": 2.999801735133512e-05, + "loss": 0.0439, + "step": 660 + }, + { + "epoch": 1.1452991452991452, + "grad_norm": 0.20380640029907227, + "learning_rate": 2.9997288332712325e-05, + "loss": 0.0424, + "step": 670 + }, + { + "epoch": 1.1623931623931625, + "grad_norm": 0.15982045233249664, + "learning_rate": 2.999644541965335e-05, + "loss": 0.0409, + "step": 680 + }, + { + "epoch": 1.1794871794871795, + "grad_norm": 0.16698749363422394, + "learning_rate": 2.9995488618559554e-05, + "loss": 0.0376, + "step": 690 + }, + { + "epoch": 1.1965811965811965, + "grad_norm": 0.13415130972862244, + "learning_rate": 2.9994417936697218e-05, + "loss": 0.0412, + "step": 700 + }, + { + "epoch": 1.2136752136752136, + "grad_norm": 0.13148538768291473, + "learning_rate": 2.9993233382197454e-05, + "loss": 0.0376, + "step": 710 + }, + { + "epoch": 1.2307692307692308, + "grad_norm": 0.16429367661476135, + "learning_rate": 2.999193496405616e-05, + "loss": 0.0398, + "step": 720 + }, + { + "epoch": 1.2478632478632479, + "grad_norm": 0.15812724828720093, + "learning_rate": 2.9990522692133962e-05, + "loss": 0.0406, + "step": 730 + }, + { + "epoch": 1.264957264957265, + "grad_norm": 0.13753320276737213, + "learning_rate": 2.9988996577156134e-05, + "loss": 0.0402, + "step": 740 + }, + { + "epoch": 1.282051282051282, + "grad_norm": 0.139229878783226, + "learning_rate": 2.9987356630712496e-05, + "loss": 0.0387, + "step": 750 + }, + { + "epoch": 1.2991452991452992, + "grad_norm": 0.15705733001232147, + "learning_rate": 2.9985602865257354e-05, + "loss": 0.0389, + "step": 760 + }, + { + "epoch": 1.3162393162393162, + "grad_norm": 0.14831461012363434, + "learning_rate": 2.998373529410939e-05, + "loss": 0.0402, + "step": 770 + }, + { + "epoch": 1.3333333333333333, + "grad_norm": 0.13418765366077423, + "learning_rate": 2.9981753931451567e-05, + "loss": 0.0385, + "step": 780 + }, + { + "epoch": 1.3504273504273505, + "grad_norm": 0.16108068823814392, + "learning_rate": 2.9979658792331016e-05, + "loss": 0.0359, + "step": 790 + }, + { + "epoch": 1.3675213675213675, + "grad_norm": 0.14137201011180878, + "learning_rate": 2.997744989265893e-05, + "loss": 0.0367, + "step": 800 + }, + { + "epoch": 1.3846153846153846, + "grad_norm": 0.15811926126480103, + "learning_rate": 2.9975127249210426e-05, + "loss": 0.0348, + "step": 810 + }, + { + "epoch": 1.4017094017094016, + "grad_norm": 0.1296013742685318, + "learning_rate": 2.9972690879624445e-05, + "loss": 0.034, + "step": 820 + }, + { + "epoch": 1.4188034188034189, + "grad_norm": 0.16444191336631775, + "learning_rate": 2.9970140802403587e-05, + "loss": 0.037, + "step": 830 + }, + { + "epoch": 1.435897435897436, + "grad_norm": 0.1809467077255249, + "learning_rate": 2.9967477036914002e-05, + "loss": 0.037, + "step": 840 + }, + { + "epoch": 1.452991452991453, + "grad_norm": 0.11857502162456512, + "learning_rate": 2.996469960338521e-05, + "loss": 0.0356, + "step": 850 + }, + { + "epoch": 1.4700854700854702, + "grad_norm": 0.12802129983901978, + "learning_rate": 2.9961808522909976e-05, + "loss": 0.0361, + "step": 860 + }, + { + "epoch": 1.4871794871794872, + "grad_norm": 0.1897210031747818, + "learning_rate": 2.995880381744414e-05, + "loss": 0.0356, + "step": 870 + }, + { + "epoch": 1.5042735042735043, + "grad_norm": 0.1568431556224823, + "learning_rate": 2.995568550980643e-05, + "loss": 0.0368, + "step": 880 + }, + { + "epoch": 1.5213675213675213, + "grad_norm": 0.14498373866081238, + "learning_rate": 2.9952453623678334e-05, + "loss": 0.0369, + "step": 890 + }, + { + "epoch": 1.5384615384615383, + "grad_norm": 0.14627397060394287, + "learning_rate": 2.9949108183603867e-05, + "loss": 0.0356, + "step": 900 + }, + { + "epoch": 1.5555555555555556, + "grad_norm": 0.143493190407753, + "learning_rate": 2.994564921498943e-05, + "loss": 0.0341, + "step": 910 + }, + { + "epoch": 1.5726495726495726, + "grad_norm": 0.1655547320842743, + "learning_rate": 2.9942076744103594e-05, + "loss": 0.0322, + "step": 920 + }, + { + "epoch": 1.5897435897435899, + "grad_norm": 0.102423295378685, + "learning_rate": 2.9938390798076886e-05, + "loss": 0.0346, + "step": 930 + }, + { + "epoch": 1.606837606837607, + "grad_norm": 0.1399497091770172, + "learning_rate": 2.993459140490163e-05, + "loss": 0.0332, + "step": 940 + }, + { + "epoch": 1.623931623931624, + "grad_norm": 0.15969839692115784, + "learning_rate": 2.9930678593431684e-05, + "loss": 0.0318, + "step": 950 + }, + { + "epoch": 1.641025641025641, + "grad_norm": 0.16585025191307068, + "learning_rate": 2.9926652393382258e-05, + "loss": 0.0326, + "step": 960 + }, + { + "epoch": 1.658119658119658, + "grad_norm": 0.13431431353092194, + "learning_rate": 2.9922512835329655e-05, + "loss": 0.0344, + "step": 970 + }, + { + "epoch": 1.6752136752136753, + "grad_norm": 0.1281099021434784, + "learning_rate": 2.9918259950711073e-05, + "loss": 0.0312, + "step": 980 + }, + { + "epoch": 1.6923076923076923, + "grad_norm": 0.13682444393634796, + "learning_rate": 2.991389377182435e-05, + "loss": 0.0338, + "step": 990 + }, + { + "epoch": 1.7094017094017095, + "grad_norm": 0.17258574068546295, + "learning_rate": 2.99094143318277e-05, + "loss": 0.03, + "step": 1000 + }, + { + "epoch": 1.7264957264957266, + "grad_norm": 0.12355158478021622, + "learning_rate": 2.9904821664739493e-05, + "loss": 0.0325, + "step": 1010 + }, + { + "epoch": 1.7435897435897436, + "grad_norm": 0.14875563979148865, + "learning_rate": 2.990011580543799e-05, + "loss": 0.0353, + "step": 1020 + }, + { + "epoch": 1.7606837606837606, + "grad_norm": 0.1353384256362915, + "learning_rate": 2.9895296789661065e-05, + "loss": 0.0294, + "step": 1030 + }, + { + "epoch": 1.7777777777777777, + "grad_norm": 0.1349417269229889, + "learning_rate": 2.989036465400594e-05, + "loss": 0.0299, + "step": 1040 + }, + { + "epoch": 1.7948717948717947, + "grad_norm": 0.17602555453777313, + "learning_rate": 2.9885319435928896e-05, + "loss": 0.028, + "step": 1050 + }, + { + "epoch": 1.811965811965812, + "grad_norm": 0.10533279925584793, + "learning_rate": 2.9880161173745017e-05, + "loss": 0.0272, + "step": 1060 + }, + { + "epoch": 1.8290598290598292, + "grad_norm": 0.12225744873285294, + "learning_rate": 2.9874889906627882e-05, + "loss": 0.0319, + "step": 1070 + }, + { + "epoch": 1.8461538461538463, + "grad_norm": 0.1326926201581955, + "learning_rate": 2.9869505674609253e-05, + "loss": 0.0281, + "step": 1080 + }, + { + "epoch": 1.8632478632478633, + "grad_norm": 0.157648503780365, + "learning_rate": 2.9864008518578794e-05, + "loss": 0.0296, + "step": 1090 + }, + { + "epoch": 1.8803418803418803, + "grad_norm": 0.191550612449646, + "learning_rate": 2.9858398480283758e-05, + "loss": 0.0311, + "step": 1100 + }, + { + "epoch": 1.8974358974358974, + "grad_norm": 0.14470480382442474, + "learning_rate": 2.9852675602328644e-05, + "loss": 0.0297, + "step": 1110 + }, + { + "epoch": 1.9145299145299144, + "grad_norm": 0.10989037901163101, + "learning_rate": 2.9846839928174923e-05, + "loss": 0.0279, + "step": 1120 + }, + { + "epoch": 1.9316239316239316, + "grad_norm": 0.12898847460746765, + "learning_rate": 2.984089150214065e-05, + "loss": 0.0324, + "step": 1130 + }, + { + "epoch": 1.9487179487179487, + "grad_norm": 0.1219201385974884, + "learning_rate": 2.9834830369400165e-05, + "loss": 0.0295, + "step": 1140 + }, + { + "epoch": 1.965811965811966, + "grad_norm": 0.14238029718399048, + "learning_rate": 2.9828656575983747e-05, + "loss": 0.0269, + "step": 1150 + }, + { + "epoch": 1.982905982905983, + "grad_norm": 0.12812891602516174, + "learning_rate": 2.9822370168777253e-05, + "loss": 0.0296, + "step": 1160 + }, + { + "epoch": 2.0, + "grad_norm": 0.16687121987342834, + "learning_rate": 2.981597119552177e-05, + "loss": 0.0319, + "step": 1170 + }, + { + "epoch": 2.017094017094017, + "grad_norm": 0.13151830434799194, + "learning_rate": 2.9809459704813236e-05, + "loss": 0.0281, + "step": 1180 + }, + { + "epoch": 2.034188034188034, + "grad_norm": 0.10349870473146439, + "learning_rate": 2.980283574610211e-05, + "loss": 0.026, + "step": 1190 + }, + { + "epoch": 2.051282051282051, + "grad_norm": 0.13649572432041168, + "learning_rate": 2.9796099369692946e-05, + "loss": 0.03, + "step": 1200 + }, + { + "epoch": 2.0683760683760686, + "grad_norm": 0.13860812783241272, + "learning_rate": 2.978925062674404e-05, + "loss": 0.0269, + "step": 1210 + }, + { + "epoch": 2.0854700854700856, + "grad_norm": 0.13259582221508026, + "learning_rate": 2.9782289569267057e-05, + "loss": 0.0275, + "step": 1220 + }, + { + "epoch": 2.1025641025641026, + "grad_norm": 0.15669459104537964, + "learning_rate": 2.977521625012659e-05, + "loss": 0.0307, + "step": 1230 + }, + { + "epoch": 2.1196581196581197, + "grad_norm": 0.1643408089876175, + "learning_rate": 2.9768030723039798e-05, + "loss": 0.0288, + "step": 1240 + }, + { + "epoch": 2.1367521367521367, + "grad_norm": 0.12998151779174805, + "learning_rate": 2.9760733042575984e-05, + "loss": 0.0282, + "step": 1250 + }, + { + "epoch": 2.1538461538461537, + "grad_norm": 0.11236439645290375, + "learning_rate": 2.9753323264156184e-05, + "loss": 0.0276, + "step": 1260 + }, + { + "epoch": 2.1709401709401708, + "grad_norm": 0.13808509707450867, + "learning_rate": 2.974580144405273e-05, + "loss": 0.0274, + "step": 1270 + }, + { + "epoch": 2.1880341880341883, + "grad_norm": 0.13739660382270813, + "learning_rate": 2.973816763938886e-05, + "loss": 0.0283, + "step": 1280 + }, + { + "epoch": 2.2051282051282053, + "grad_norm": 0.13221029937267303, + "learning_rate": 2.9730421908138236e-05, + "loss": 0.0279, + "step": 1290 + }, + { + "epoch": 2.2222222222222223, + "grad_norm": 0.1287679374217987, + "learning_rate": 2.9722564309124546e-05, + "loss": 0.029, + "step": 1300 + }, + { + "epoch": 2.2393162393162394, + "grad_norm": 0.12047926336526871, + "learning_rate": 2.971459490202102e-05, + "loss": 0.0268, + "step": 1310 + }, + { + "epoch": 2.2564102564102564, + "grad_norm": 0.09681694209575653, + "learning_rate": 2.9706513747350017e-05, + "loss": 0.0254, + "step": 1320 + }, + { + "epoch": 2.2735042735042734, + "grad_norm": 0.14194591343402863, + "learning_rate": 2.9698320906482535e-05, + "loss": 0.0295, + "step": 1330 + }, + { + "epoch": 2.2905982905982905, + "grad_norm": 0.12847678363323212, + "learning_rate": 2.9690016441637747e-05, + "loss": 0.0269, + "step": 1340 + }, + { + "epoch": 2.3076923076923075, + "grad_norm": 0.15464898943901062, + "learning_rate": 2.9681600415882556e-05, + "loss": 0.0269, + "step": 1350 + }, + { + "epoch": 2.324786324786325, + "grad_norm": 0.12396308779716492, + "learning_rate": 2.9673072893131068e-05, + "loss": 0.0286, + "step": 1360 + }, + { + "epoch": 2.341880341880342, + "grad_norm": 0.14894050359725952, + "learning_rate": 2.9664433938144162e-05, + "loss": 0.0288, + "step": 1370 + }, + { + "epoch": 2.358974358974359, + "grad_norm": 0.13231822848320007, + "learning_rate": 2.9655683616528958e-05, + "loss": 0.0266, + "step": 1380 + }, + { + "epoch": 2.376068376068376, + "grad_norm": 0.145207017660141, + "learning_rate": 2.9646821994738322e-05, + "loss": 0.0257, + "step": 1390 + }, + { + "epoch": 2.393162393162393, + "grad_norm": 0.14421381056308746, + "learning_rate": 2.9637849140070393e-05, + "loss": 0.0257, + "step": 1400 + }, + { + "epoch": 2.41025641025641, + "grad_norm": 0.1812671273946762, + "learning_rate": 2.9628765120668033e-05, + "loss": 0.0256, + "step": 1410 + }, + { + "epoch": 2.427350427350427, + "grad_norm": 0.1101057231426239, + "learning_rate": 2.9619570005518336e-05, + "loss": 0.0263, + "step": 1420 + }, + { + "epoch": 2.4444444444444446, + "grad_norm": 0.12395047396421432, + "learning_rate": 2.9610263864452077e-05, + "loss": 0.025, + "step": 1430 + }, + { + "epoch": 2.4615384615384617, + "grad_norm": 0.1621936559677124, + "learning_rate": 2.9600846768143225e-05, + "loss": 0.0261, + "step": 1440 + }, + { + "epoch": 2.4786324786324787, + "grad_norm": 0.13965119421482086, + "learning_rate": 2.959131878810836e-05, + "loss": 0.0251, + "step": 1450 + }, + { + "epoch": 2.4957264957264957, + "grad_norm": 0.11430062353610992, + "learning_rate": 2.9581679996706157e-05, + "loss": 0.0263, + "step": 1460 + }, + { + "epoch": 2.5128205128205128, + "grad_norm": 0.1259123533964157, + "learning_rate": 2.9571930467136833e-05, + "loss": 0.0246, + "step": 1470 + }, + { + "epoch": 2.52991452991453, + "grad_norm": 0.13637065887451172, + "learning_rate": 2.956207027344159e-05, + "loss": 0.0253, + "step": 1480 + }, + { + "epoch": 2.547008547008547, + "grad_norm": 0.1521414965391159, + "learning_rate": 2.9552099490502026e-05, + "loss": 0.0248, + "step": 1490 + }, + { + "epoch": 2.564102564102564, + "grad_norm": 0.17049413919448853, + "learning_rate": 2.9542018194039634e-05, + "loss": 0.0262, + "step": 1500 + }, + { + "epoch": 2.5811965811965814, + "grad_norm": 0.09336340427398682, + "learning_rate": 2.953182646061514e-05, + "loss": 0.0263, + "step": 1510 + }, + { + "epoch": 2.5982905982905984, + "grad_norm": 0.12669335305690765, + "learning_rate": 2.9521524367627994e-05, + "loss": 0.0265, + "step": 1520 + }, + { + "epoch": 2.6153846153846154, + "grad_norm": 0.12848782539367676, + "learning_rate": 2.9511111993315743e-05, + "loss": 0.026, + "step": 1530 + }, + { + "epoch": 2.6324786324786325, + "grad_norm": 0.14347173273563385, + "learning_rate": 2.9500589416753447e-05, + "loss": 0.0244, + "step": 1540 + }, + { + "epoch": 2.6495726495726495, + "grad_norm": 0.10938625037670135, + "learning_rate": 2.9489956717853088e-05, + "loss": 0.0243, + "step": 1550 + }, + { + "epoch": 2.6666666666666665, + "grad_norm": 0.11029652506113052, + "learning_rate": 2.9479213977362935e-05, + "loss": 0.0277, + "step": 1560 + }, + { + "epoch": 2.683760683760684, + "grad_norm": 0.14064879715442657, + "learning_rate": 2.9468361276866973e-05, + "loss": 0.0257, + "step": 1570 + }, + { + "epoch": 2.700854700854701, + "grad_norm": 0.13628490269184113, + "learning_rate": 2.9457398698784243e-05, + "loss": 0.0261, + "step": 1580 + }, + { + "epoch": 2.717948717948718, + "grad_norm": 0.10640981048345566, + "learning_rate": 2.944632632636823e-05, + "loss": 0.0236, + "step": 1590 + }, + { + "epoch": 2.735042735042735, + "grad_norm": 0.10401084274053574, + "learning_rate": 2.9435144243706255e-05, + "loss": 0.0262, + "step": 1600 + }, + { + "epoch": 2.752136752136752, + "grad_norm": 0.11106079071760178, + "learning_rate": 2.9423852535718805e-05, + "loss": 0.0249, + "step": 1610 + }, + { + "epoch": 2.769230769230769, + "grad_norm": 0.18114276230335236, + "learning_rate": 2.9412451288158883e-05, + "loss": 0.024, + "step": 1620 + }, + { + "epoch": 2.786324786324786, + "grad_norm": 0.1303853988647461, + "learning_rate": 2.94009405876114e-05, + "loss": 0.0239, + "step": 1630 + }, + { + "epoch": 2.8034188034188032, + "grad_norm": 0.11971442401409149, + "learning_rate": 2.9389320521492464e-05, + "loss": 0.023, + "step": 1640 + }, + { + "epoch": 2.8205128205128203, + "grad_norm": 0.1276145875453949, + "learning_rate": 2.9377591178048765e-05, + "loss": 0.0237, + "step": 1650 + }, + { + "epoch": 2.8376068376068377, + "grad_norm": 0.1473803073167801, + "learning_rate": 2.9365752646356857e-05, + "loss": 0.0232, + "step": 1660 + }, + { + "epoch": 2.8547008547008548, + "grad_norm": 0.14460980892181396, + "learning_rate": 2.935380501632253e-05, + "loss": 0.0236, + "step": 1670 + }, + { + "epoch": 2.871794871794872, + "grad_norm": 0.1054682806134224, + "learning_rate": 2.934174837868008e-05, + "loss": 0.0237, + "step": 1680 + }, + { + "epoch": 2.888888888888889, + "grad_norm": 0.09902238100767136, + "learning_rate": 2.932958282499167e-05, + "loss": 0.0262, + "step": 1690 + }, + { + "epoch": 2.905982905982906, + "grad_norm": 0.14733590185642242, + "learning_rate": 2.9317308447646585e-05, + "loss": 0.0275, + "step": 1700 + }, + { + "epoch": 2.9230769230769234, + "grad_norm": 0.1391238123178482, + "learning_rate": 2.9304925339860565e-05, + "loss": 0.0254, + "step": 1710 + }, + { + "epoch": 2.9401709401709404, + "grad_norm": 0.1650138944387436, + "learning_rate": 2.929243359567509e-05, + "loss": 0.0228, + "step": 1720 + }, + { + "epoch": 2.9572649572649574, + "grad_norm": 0.13160136342048645, + "learning_rate": 2.9279833309956637e-05, + "loss": 0.0245, + "step": 1730 + }, + { + "epoch": 2.9743589743589745, + "grad_norm": 0.1312543749809265, + "learning_rate": 2.9267124578396013e-05, + "loss": 0.0221, + "step": 1740 + }, + { + "epoch": 2.9914529914529915, + "grad_norm": 0.13742756843566895, + "learning_rate": 2.9254307497507584e-05, + "loss": 0.0256, + "step": 1750 + }, + { + "epoch": 3.0085470085470085, + "grad_norm": 0.08947888761758804, + "learning_rate": 2.924138216462856e-05, + "loss": 0.0247, + "step": 1760 + }, + { + "epoch": 3.0256410256410255, + "grad_norm": 0.09200987219810486, + "learning_rate": 2.922834867791825e-05, + "loss": 0.0223, + "step": 1770 + }, + { + "epoch": 3.0427350427350426, + "grad_norm": 0.14834287762641907, + "learning_rate": 2.9215207136357322e-05, + "loss": 0.0254, + "step": 1780 + }, + { + "epoch": 3.0598290598290596, + "grad_norm": 0.13650742173194885, + "learning_rate": 2.9201957639747056e-05, + "loss": 0.0248, + "step": 1790 + }, + { + "epoch": 3.076923076923077, + "grad_norm": 0.10533406585454941, + "learning_rate": 2.9188600288708565e-05, + "loss": 0.0239, + "step": 1800 + }, + { + "epoch": 3.094017094017094, + "grad_norm": 0.09621796756982803, + "learning_rate": 2.9175135184682053e-05, + "loss": 0.0219, + "step": 1810 + }, + { + "epoch": 3.111111111111111, + "grad_norm": 0.1239803209900856, + "learning_rate": 2.9161562429926038e-05, + "loss": 0.0228, + "step": 1820 + }, + { + "epoch": 3.128205128205128, + "grad_norm": 0.13734465837478638, + "learning_rate": 2.9147882127516568e-05, + "loss": 0.0229, + "step": 1830 + }, + { + "epoch": 3.1452991452991452, + "grad_norm": 0.11425598710775375, + "learning_rate": 2.913409438134644e-05, + "loss": 0.0224, + "step": 1840 + }, + { + "epoch": 3.1623931623931623, + "grad_norm": 0.16756656765937805, + "learning_rate": 2.9120199296124436e-05, + "loss": 0.025, + "step": 1850 + }, + { + "epoch": 3.1794871794871793, + "grad_norm": 0.11103396117687225, + "learning_rate": 2.9106196977374477e-05, + "loss": 0.0236, + "step": 1860 + }, + { + "epoch": 3.1965811965811968, + "grad_norm": 0.10024338215589523, + "learning_rate": 2.9092087531434878e-05, + "loss": 0.0219, + "step": 1870 + }, + { + "epoch": 3.213675213675214, + "grad_norm": 0.11945660412311554, + "learning_rate": 2.9077871065457495e-05, + "loss": 0.0239, + "step": 1880 + }, + { + "epoch": 3.230769230769231, + "grad_norm": 0.11576039344072342, + "learning_rate": 2.9063547687406945e-05, + "loss": 0.0243, + "step": 1890 + }, + { + "epoch": 3.247863247863248, + "grad_norm": 0.09603351354598999, + "learning_rate": 2.9049117506059756e-05, + "loss": 0.0228, + "step": 1900 + }, + { + "epoch": 3.264957264957265, + "grad_norm": 0.07817462831735611, + "learning_rate": 2.9034580631003564e-05, + "loss": 0.0235, + "step": 1910 + }, + { + "epoch": 3.282051282051282, + "grad_norm": 0.11091442406177521, + "learning_rate": 2.9019937172636276e-05, + "loss": 0.0237, + "step": 1920 + }, + { + "epoch": 3.299145299145299, + "grad_norm": 0.13915278017520905, + "learning_rate": 2.9005187242165228e-05, + "loss": 0.0228, + "step": 1930 + }, + { + "epoch": 3.316239316239316, + "grad_norm": 0.09350527077913284, + "learning_rate": 2.899033095160633e-05, + "loss": 0.0231, + "step": 1940 + }, + { + "epoch": 3.3333333333333335, + "grad_norm": 0.11766257882118225, + "learning_rate": 2.8975368413783235e-05, + "loss": 0.0213, + "step": 1950 + }, + { + "epoch": 3.3504273504273505, + "grad_norm": 0.0914580225944519, + "learning_rate": 2.8960299742326473e-05, + "loss": 0.0233, + "step": 1960 + }, + { + "epoch": 3.3675213675213675, + "grad_norm": 0.12332584708929062, + "learning_rate": 2.8945125051672595e-05, + "loss": 0.0245, + "step": 1970 + }, + { + "epoch": 3.3846153846153846, + "grad_norm": 0.10972987115383148, + "learning_rate": 2.892984445706328e-05, + "loss": 0.0234, + "step": 1980 + }, + { + "epoch": 3.4017094017094016, + "grad_norm": 0.1142164021730423, + "learning_rate": 2.891445807454449e-05, + "loss": 0.0235, + "step": 1990 + }, + { + "epoch": 3.4188034188034186, + "grad_norm": 0.13368234038352966, + "learning_rate": 2.8898966020965576e-05, + "loss": 0.0224, + "step": 2000 + }, + { + "epoch": 3.435897435897436, + "grad_norm": 0.10849815607070923, + "learning_rate": 2.8883368413978386e-05, + "loss": 0.0213, + "step": 2010 + }, + { + "epoch": 3.452991452991453, + "grad_norm": 0.09034600108861923, + "learning_rate": 2.8867665372036375e-05, + "loss": 0.0239, + "step": 2020 + }, + { + "epoch": 3.47008547008547, + "grad_norm": 0.12289014458656311, + "learning_rate": 2.8851857014393705e-05, + "loss": 0.022, + "step": 2030 + }, + { + "epoch": 3.4871794871794872, + "grad_norm": 0.15092331171035767, + "learning_rate": 2.8835943461104354e-05, + "loss": 0.0212, + "step": 2040 + }, + { + "epoch": 3.5042735042735043, + "grad_norm": 0.11829588562250137, + "learning_rate": 2.8819924833021176e-05, + "loss": 0.0235, + "step": 2050 + }, + { + "epoch": 3.5213675213675213, + "grad_norm": 0.11106596887111664, + "learning_rate": 2.8803801251795007e-05, + "loss": 0.0223, + "step": 2060 + }, + { + "epoch": 3.5384615384615383, + "grad_norm": 0.0992603749036789, + "learning_rate": 2.8787572839873725e-05, + "loss": 0.024, + "step": 2070 + }, + { + "epoch": 3.5555555555555554, + "grad_norm": 0.11236703395843506, + "learning_rate": 2.8771239720501333e-05, + "loss": 0.0203, + "step": 2080 + }, + { + "epoch": 3.5726495726495724, + "grad_norm": 0.14634686708450317, + "learning_rate": 2.8754802017717015e-05, + "loss": 0.0244, + "step": 2090 + }, + { + "epoch": 3.58974358974359, + "grad_norm": 0.09683041274547577, + "learning_rate": 2.87382598563542e-05, + "loss": 0.0227, + "step": 2100 + }, + { + "epoch": 3.606837606837607, + "grad_norm": 0.13424035906791687, + "learning_rate": 2.8721613362039607e-05, + "loss": 0.0215, + "step": 2110 + }, + { + "epoch": 3.623931623931624, + "grad_norm": 0.10873326659202576, + "learning_rate": 2.8704862661192295e-05, + "loss": 0.0215, + "step": 2120 + }, + { + "epoch": 3.641025641025641, + "grad_norm": 0.0864587277173996, + "learning_rate": 2.86880078810227e-05, + "loss": 0.0251, + "step": 2130 + }, + { + "epoch": 3.658119658119658, + "grad_norm": 0.11012615263462067, + "learning_rate": 2.8671049149531677e-05, + "loss": 0.0199, + "step": 2140 + }, + { + "epoch": 3.6752136752136755, + "grad_norm": 0.18411065638065338, + "learning_rate": 2.8653986595509524e-05, + "loss": 0.0232, + "step": 2150 + }, + { + "epoch": 3.6923076923076925, + "grad_norm": 0.10877490788698196, + "learning_rate": 2.8636820348534987e-05, + "loss": 0.0207, + "step": 2160 + }, + { + "epoch": 3.7094017094017095, + "grad_norm": 0.10099867731332779, + "learning_rate": 2.861955053897431e-05, + "loss": 0.0175, + "step": 2170 + }, + { + "epoch": 3.7264957264957266, + "grad_norm": 0.08751079440116882, + "learning_rate": 2.8602177297980213e-05, + "loss": 0.0212, + "step": 2180 + }, + { + "epoch": 3.7435897435897436, + "grad_norm": 0.13080573081970215, + "learning_rate": 2.8584700757490913e-05, + "loss": 0.0208, + "step": 2190 + }, + { + "epoch": 3.7606837606837606, + "grad_norm": 0.0996410921216011, + "learning_rate": 2.8567121050229122e-05, + "loss": 0.024, + "step": 2200 + }, + { + "epoch": 3.7777777777777777, + "grad_norm": 0.12667258083820343, + "learning_rate": 2.8549438309701028e-05, + "loss": 0.0227, + "step": 2210 + }, + { + "epoch": 3.7948717948717947, + "grad_norm": 0.12759651243686676, + "learning_rate": 2.8531652670195292e-05, + "loss": 0.0215, + "step": 2220 + }, + { + "epoch": 3.8119658119658117, + "grad_norm": 0.10755474865436554, + "learning_rate": 2.8513764266782025e-05, + "loss": 0.0226, + "step": 2230 + }, + { + "epoch": 3.8290598290598292, + "grad_norm": 0.1080416887998581, + "learning_rate": 2.849577323531176e-05, + "loss": 0.021, + "step": 2240 + }, + { + "epoch": 3.8461538461538463, + "grad_norm": 0.09134095907211304, + "learning_rate": 2.8477679712414424e-05, + "loss": 0.0213, + "step": 2250 + }, + { + "epoch": 3.8632478632478633, + "grad_norm": 0.09035451710224152, + "learning_rate": 2.8459483835498298e-05, + "loss": 0.0227, + "step": 2260 + }, + { + "epoch": 3.8803418803418803, + "grad_norm": 0.11327807605266571, + "learning_rate": 2.844118574274896e-05, + "loss": 0.0211, + "step": 2270 + }, + { + "epoch": 3.8974358974358974, + "grad_norm": 0.118441142141819, + "learning_rate": 2.842278557312828e-05, + "loss": 0.0191, + "step": 2280 + }, + { + "epoch": 3.9145299145299144, + "grad_norm": 0.0991198942065239, + "learning_rate": 2.84042834663733e-05, + "loss": 0.0192, + "step": 2290 + }, + { + "epoch": 3.931623931623932, + "grad_norm": 0.08809023350477219, + "learning_rate": 2.8385679562995235e-05, + "loss": 0.0202, + "step": 2300 + }, + { + "epoch": 3.948717948717949, + "grad_norm": 0.10185550898313522, + "learning_rate": 2.8366974004278357e-05, + "loss": 0.0232, + "step": 2310 + }, + { + "epoch": 3.965811965811966, + "grad_norm": 0.11275793612003326, + "learning_rate": 2.834816693227895e-05, + "loss": 0.0197, + "step": 2320 + }, + { + "epoch": 3.982905982905983, + "grad_norm": 0.1458314061164856, + "learning_rate": 2.8329258489824234e-05, + "loss": 0.0246, + "step": 2330 + }, + { + "epoch": 4.0, + "grad_norm": 0.08866702765226364, + "learning_rate": 2.8310248820511248e-05, + "loss": 0.0196, + "step": 2340 + }, + { + "epoch": 4.017094017094017, + "grad_norm": 0.1087232232093811, + "learning_rate": 2.829113806870581e-05, + "loss": 0.0222, + "step": 2350 + }, + { + "epoch": 4.034188034188034, + "grad_norm": 0.13110089302062988, + "learning_rate": 2.8271926379541363e-05, + "loss": 0.0213, + "step": 2360 + }, + { + "epoch": 4.051282051282051, + "grad_norm": 0.07700895518064499, + "learning_rate": 2.8252613898917923e-05, + "loss": 0.02, + "step": 2370 + }, + { + "epoch": 4.068376068376068, + "grad_norm": 0.09605945646762848, + "learning_rate": 2.8233200773500942e-05, + "loss": 0.021, + "step": 2380 + }, + { + "epoch": 4.085470085470085, + "grad_norm": 0.11480522155761719, + "learning_rate": 2.8213687150720204e-05, + "loss": 0.0223, + "step": 2390 + }, + { + "epoch": 4.102564102564102, + "grad_norm": 0.12132875621318817, + "learning_rate": 2.8194073178768703e-05, + "loss": 0.0192, + "step": 2400 + }, + { + "epoch": 4.119658119658119, + "grad_norm": 0.09794069826602936, + "learning_rate": 2.817435900660152e-05, + "loss": 0.0189, + "step": 2410 + }, + { + "epoch": 4.136752136752137, + "grad_norm": 0.09901175647974014, + "learning_rate": 2.8154544783934688e-05, + "loss": 0.0238, + "step": 2420 + }, + { + "epoch": 4.153846153846154, + "grad_norm": 0.10227429866790771, + "learning_rate": 2.8134630661244055e-05, + "loss": 0.0204, + "step": 2430 + }, + { + "epoch": 4.170940170940171, + "grad_norm": 0.12511932849884033, + "learning_rate": 2.8114616789764146e-05, + "loss": 0.0209, + "step": 2440 + }, + { + "epoch": 4.188034188034188, + "grad_norm": 0.10401079803705215, + "learning_rate": 2.8094503321487017e-05, + "loss": 0.02, + "step": 2450 + }, + { + "epoch": 4.205128205128205, + "grad_norm": 0.09629535675048828, + "learning_rate": 2.8074290409161084e-05, + "loss": 0.0195, + "step": 2460 + }, + { + "epoch": 4.222222222222222, + "grad_norm": 0.0992991030216217, + "learning_rate": 2.8053978206289984e-05, + "loss": 0.0191, + "step": 2470 + }, + { + "epoch": 4.239316239316239, + "grad_norm": 0.0984906479716301, + "learning_rate": 2.803356686713139e-05, + "loss": 0.0191, + "step": 2480 + }, + { + "epoch": 4.256410256410256, + "grad_norm": 0.10978572815656662, + "learning_rate": 2.801305654669587e-05, + "loss": 0.0223, + "step": 2490 + }, + { + "epoch": 4.273504273504273, + "grad_norm": 0.09285573661327362, + "learning_rate": 2.799244740074566e-05, + "loss": 0.0195, + "step": 2500 + }, + { + "epoch": 4.2905982905982905, + "grad_norm": 0.08738566935062408, + "learning_rate": 2.797173958579353e-05, + "loss": 0.0228, + "step": 2510 + }, + { + "epoch": 4.3076923076923075, + "grad_norm": 0.11125430464744568, + "learning_rate": 2.7950933259101575e-05, + "loss": 0.0203, + "step": 2520 + }, + { + "epoch": 4.3247863247863245, + "grad_norm": 0.10201579332351685, + "learning_rate": 2.793002857868001e-05, + "loss": 0.0214, + "step": 2530 + }, + { + "epoch": 4.3418803418803416, + "grad_norm": 0.10354452580213547, + "learning_rate": 2.7909025703285983e-05, + "loss": 0.0199, + "step": 2540 + }, + { + "epoch": 4.358974358974359, + "grad_norm": 0.09892477840185165, + "learning_rate": 2.7887924792422385e-05, + "loss": 0.0224, + "step": 2550 + }, + { + "epoch": 4.3760683760683765, + "grad_norm": 0.1194918155670166, + "learning_rate": 2.78667260063366e-05, + "loss": 0.0231, + "step": 2560 + }, + { + "epoch": 4.3931623931623935, + "grad_norm": 0.11821082979440689, + "learning_rate": 2.784542950601931e-05, + "loss": 0.0196, + "step": 2570 + }, + { + "epoch": 4.410256410256411, + "grad_norm": 0.09250211715698242, + "learning_rate": 2.7824035453203293e-05, + "loss": 0.0203, + "step": 2580 + }, + { + "epoch": 4.427350427350428, + "grad_norm": 0.12260550260543823, + "learning_rate": 2.7802544010362154e-05, + "loss": 0.0186, + "step": 2590 + }, + { + "epoch": 4.444444444444445, + "grad_norm": 0.0743408128619194, + "learning_rate": 2.778095534070912e-05, + "loss": 0.0209, + "step": 2600 + }, + { + "epoch": 4.461538461538462, + "grad_norm": 0.08710762113332748, + "learning_rate": 2.7759269608195777e-05, + "loss": 0.0204, + "step": 2610 + }, + { + "epoch": 4.478632478632479, + "grad_norm": 0.08449119329452515, + "learning_rate": 2.773748697751086e-05, + "loss": 0.0199, + "step": 2620 + }, + { + "epoch": 4.495726495726496, + "grad_norm": 0.09476049244403839, + "learning_rate": 2.7715607614078956e-05, + "loss": 0.0201, + "step": 2630 + }, + { + "epoch": 4.512820512820513, + "grad_norm": 0.09813626855611801, + "learning_rate": 2.7693631684059305e-05, + "loss": 0.0204, + "step": 2640 + }, + { + "epoch": 4.52991452991453, + "grad_norm": 0.10010302066802979, + "learning_rate": 2.767155935434448e-05, + "loss": 0.0194, + "step": 2650 + }, + { + "epoch": 4.547008547008547, + "grad_norm": 0.1183779239654541, + "learning_rate": 2.7649390792559146e-05, + "loss": 0.0203, + "step": 2660 + }, + { + "epoch": 4.564102564102564, + "grad_norm": 0.08748367428779602, + "learning_rate": 2.7627126167058807e-05, + "loss": 0.0202, + "step": 2670 + }, + { + "epoch": 4.581196581196581, + "grad_norm": 0.12158568203449249, + "learning_rate": 2.7604765646928482e-05, + "loss": 0.0207, + "step": 2680 + }, + { + "epoch": 4.598290598290598, + "grad_norm": 0.1865331381559372, + "learning_rate": 2.758230940198147e-05, + "loss": 0.0208, + "step": 2690 + }, + { + "epoch": 4.615384615384615, + "grad_norm": 0.1120152547955513, + "learning_rate": 2.755975760275802e-05, + "loss": 0.0199, + "step": 2700 + }, + { + "epoch": 4.632478632478632, + "grad_norm": 0.08776689320802689, + "learning_rate": 2.7537110420524057e-05, + "loss": 0.0193, + "step": 2710 + }, + { + "epoch": 4.64957264957265, + "grad_norm": 0.13004325330257416, + "learning_rate": 2.7514368027269875e-05, + "loss": 0.0187, + "step": 2720 + }, + { + "epoch": 4.666666666666667, + "grad_norm": 0.09902483224868774, + "learning_rate": 2.749153059570883e-05, + "loss": 0.0201, + "step": 2730 + }, + { + "epoch": 4.683760683760684, + "grad_norm": 0.1488153040409088, + "learning_rate": 2.7468598299276044e-05, + "loss": 0.0191, + "step": 2740 + }, + { + "epoch": 4.700854700854701, + "grad_norm": 0.10970397293567657, + "learning_rate": 2.7445571312127058e-05, + "loss": 0.0184, + "step": 2750 + }, + { + "epoch": 4.717948717948718, + "grad_norm": 0.09405538439750671, + "learning_rate": 2.7422449809136538e-05, + "loss": 0.0195, + "step": 2760 + }, + { + "epoch": 4.735042735042735, + "grad_norm": 0.09082396328449249, + "learning_rate": 2.739923396589693e-05, + "loss": 0.0204, + "step": 2770 + }, + { + "epoch": 4.752136752136752, + "grad_norm": 0.1001177504658699, + "learning_rate": 2.737592395871713e-05, + "loss": 0.0173, + "step": 2780 + }, + { + "epoch": 4.769230769230769, + "grad_norm": 0.09204474091529846, + "learning_rate": 2.7352519964621146e-05, + "loss": 0.0185, + "step": 2790 + }, + { + "epoch": 4.786324786324786, + "grad_norm": 0.09500245749950409, + "learning_rate": 2.7329022161346758e-05, + "loss": 0.0198, + "step": 2800 + }, + { + "epoch": 4.803418803418803, + "grad_norm": 0.08855962008237839, + "learning_rate": 2.730543072734417e-05, + "loss": 0.0209, + "step": 2810 + }, + { + "epoch": 4.82051282051282, + "grad_norm": 0.09660346806049347, + "learning_rate": 2.728174584177463e-05, + "loss": 0.0197, + "step": 2820 + }, + { + "epoch": 4.837606837606837, + "grad_norm": 0.11314470320940018, + "learning_rate": 2.725796768450911e-05, + "loss": 0.0181, + "step": 2830 + }, + { + "epoch": 4.854700854700854, + "grad_norm": 0.09389517456293106, + "learning_rate": 2.7234096436126903e-05, + "loss": 0.02, + "step": 2840 + }, + { + "epoch": 4.871794871794872, + "grad_norm": 0.09539874643087387, + "learning_rate": 2.721013227791428e-05, + "loss": 0.0209, + "step": 2850 + }, + { + "epoch": 4.888888888888889, + "grad_norm": 0.09690243750810623, + "learning_rate": 2.7186075391863088e-05, + "loss": 0.0194, + "step": 2860 + }, + { + "epoch": 4.905982905982906, + "grad_norm": 0.10651098191738129, + "learning_rate": 2.7161925960669383e-05, + "loss": 0.0219, + "step": 2870 + }, + { + "epoch": 4.923076923076923, + "grad_norm": 0.12198858708143234, + "learning_rate": 2.713768416773205e-05, + "loss": 0.0194, + "step": 2880 + }, + { + "epoch": 4.94017094017094, + "grad_norm": 0.0976480171084404, + "learning_rate": 2.7113350197151394e-05, + "loss": 0.0203, + "step": 2890 + }, + { + "epoch": 4.957264957264957, + "grad_norm": 0.10814037919044495, + "learning_rate": 2.708892423372774e-05, + "loss": 0.0208, + "step": 2900 + }, + { + "epoch": 4.9743589743589745, + "grad_norm": 0.08893640339374542, + "learning_rate": 2.7064406462960044e-05, + "loss": 0.0167, + "step": 2910 + }, + { + "epoch": 4.9914529914529915, + "grad_norm": 0.1479610800743103, + "learning_rate": 2.7039797071044482e-05, + "loss": 0.022, + "step": 2920 + }, + { + "epoch": 5.0085470085470085, + "grad_norm": 0.09371191263198853, + "learning_rate": 2.7015096244873028e-05, + "loss": 0.0199, + "step": 2930 + }, + { + "epoch": 5.0256410256410255, + "grad_norm": 0.12326940894126892, + "learning_rate": 2.6990304172032038e-05, + "loss": 0.0195, + "step": 2940 + }, + { + "epoch": 5.042735042735043, + "grad_norm": 0.1065201461315155, + "learning_rate": 2.6965421040800823e-05, + "loss": 0.02, + "step": 2950 + }, + { + "epoch": 5.05982905982906, + "grad_norm": 0.059518881142139435, + "learning_rate": 2.6940447040150228e-05, + "loss": 0.0183, + "step": 2960 + }, + { + "epoch": 5.076923076923077, + "grad_norm": 0.1376444697380066, + "learning_rate": 2.691538235974118e-05, + "loss": 0.0196, + "step": 2970 + }, + { + "epoch": 5.094017094017094, + "grad_norm": 0.09283936023712158, + "learning_rate": 2.6890227189923285e-05, + "loss": 0.0176, + "step": 2980 + }, + { + "epoch": 5.111111111111111, + "grad_norm": 0.10842864215373993, + "learning_rate": 2.6864981721733317e-05, + "loss": 0.0177, + "step": 2990 + }, + { + "epoch": 5.128205128205128, + "grad_norm": 0.10565061122179031, + "learning_rate": 2.683964614689383e-05, + "loss": 0.0188, + "step": 3000 + }, + { + "epoch": 5.145299145299146, + "grad_norm": 0.09310047328472137, + "learning_rate": 2.681422065781169e-05, + "loss": 0.0171, + "step": 3010 + }, + { + "epoch": 5.162393162393163, + "grad_norm": 0.10931243002414703, + "learning_rate": 2.6788705447576563e-05, + "loss": 0.0186, + "step": 3020 + }, + { + "epoch": 5.17948717948718, + "grad_norm": 0.082731693983078, + "learning_rate": 2.6763100709959524e-05, + "loss": 0.0164, + "step": 3030 + }, + { + "epoch": 5.196581196581197, + "grad_norm": 0.13504242897033691, + "learning_rate": 2.6737406639411537e-05, + "loss": 0.0194, + "step": 3040 + }, + { + "epoch": 5.213675213675214, + "grad_norm": 0.0828668624162674, + "learning_rate": 2.6711623431061985e-05, + "loss": 0.0159, + "step": 3050 + }, + { + "epoch": 5.230769230769231, + "grad_norm": 0.10685226321220398, + "learning_rate": 2.6685751280717195e-05, + "loss": 0.0173, + "step": 3060 + }, + { + "epoch": 5.247863247863248, + "grad_norm": 0.09575466811656952, + "learning_rate": 2.665979038485896e-05, + "loss": 0.0182, + "step": 3070 + }, + { + "epoch": 5.264957264957265, + "grad_norm": 0.09520144015550613, + "learning_rate": 2.6633740940643022e-05, + "loss": 0.0183, + "step": 3080 + }, + { + "epoch": 5.282051282051282, + "grad_norm": 0.07664638757705688, + "learning_rate": 2.6607603145897594e-05, + "loss": 0.019, + "step": 3090 + }, + { + "epoch": 5.299145299145299, + "grad_norm": 0.10608942806720734, + "learning_rate": 2.6581377199121853e-05, + "loss": 0.0221, + "step": 3100 + }, + { + "epoch": 5.316239316239316, + "grad_norm": 0.08229487389326096, + "learning_rate": 2.6555063299484434e-05, + "loss": 0.0176, + "step": 3110 + }, + { + "epoch": 5.333333333333333, + "grad_norm": 0.0870119258761406, + "learning_rate": 2.6528661646821917e-05, + "loss": 0.0177, + "step": 3120 + }, + { + "epoch": 5.35042735042735, + "grad_norm": 0.13353557884693146, + "learning_rate": 2.6502172441637302e-05, + "loss": 0.0188, + "step": 3130 + }, + { + "epoch": 5.367521367521368, + "grad_norm": 0.09941720962524414, + "learning_rate": 2.6475595885098493e-05, + "loss": 0.0183, + "step": 3140 + }, + { + "epoch": 5.384615384615385, + "grad_norm": 0.09147895872592926, + "learning_rate": 2.644893217903677e-05, + "loss": 0.0195, + "step": 3150 + }, + { + "epoch": 5.401709401709402, + "grad_norm": 0.08084908127784729, + "learning_rate": 2.6422181525945265e-05, + "loss": 0.0164, + "step": 3160 + }, + { + "epoch": 5.418803418803419, + "grad_norm": 0.09985609352588654, + "learning_rate": 2.63953441289774e-05, + "loss": 0.0185, + "step": 3170 + }, + { + "epoch": 5.435897435897436, + "grad_norm": 0.10765116661787033, + "learning_rate": 2.6368420191945365e-05, + "loss": 0.0175, + "step": 3180 + }, + { + "epoch": 5.452991452991453, + "grad_norm": 0.07902234047651291, + "learning_rate": 2.6341409919318572e-05, + "loss": 0.0176, + "step": 3190 + }, + { + "epoch": 5.47008547008547, + "grad_norm": 0.08553142845630646, + "learning_rate": 2.6314313516222084e-05, + "loss": 0.0197, + "step": 3200 + }, + { + "epoch": 5.487179487179487, + "grad_norm": 0.10211332142353058, + "learning_rate": 2.6287131188435075e-05, + "loss": 0.0192, + "step": 3210 + }, + { + "epoch": 5.504273504273504, + "grad_norm": 0.09006617218255997, + "learning_rate": 2.6259863142389252e-05, + "loss": 0.0164, + "step": 3220 + }, + { + "epoch": 5.521367521367521, + "grad_norm": 0.09639620035886765, + "learning_rate": 2.62325095851673e-05, + "loss": 0.0151, + "step": 3230 + }, + { + "epoch": 5.538461538461538, + "grad_norm": 0.07281971722841263, + "learning_rate": 2.620507072450131e-05, + "loss": 0.0183, + "step": 3240 + }, + { + "epoch": 5.555555555555555, + "grad_norm": 0.06957155466079712, + "learning_rate": 2.6177546768771183e-05, + "loss": 0.0172, + "step": 3250 + }, + { + "epoch": 5.572649572649572, + "grad_norm": 0.08119498938322067, + "learning_rate": 2.6149937927003062e-05, + "loss": 0.019, + "step": 3260 + }, + { + "epoch": 5.589743589743589, + "grad_norm": 0.07163924723863602, + "learning_rate": 2.6122244408867766e-05, + "loss": 0.0194, + "step": 3270 + }, + { + "epoch": 5.6068376068376065, + "grad_norm": 0.09349033981561661, + "learning_rate": 2.6094466424679148e-05, + "loss": 0.0189, + "step": 3280 + }, + { + "epoch": 5.6239316239316235, + "grad_norm": 0.08922473341226578, + "learning_rate": 2.606660418539253e-05, + "loss": 0.0215, + "step": 3290 + }, + { + "epoch": 5.641025641025641, + "grad_norm": 0.10527209937572479, + "learning_rate": 2.6038657902603107e-05, + "loss": 0.018, + "step": 3300 + }, + { + "epoch": 5.6581196581196584, + "grad_norm": 0.1067451685667038, + "learning_rate": 2.6010627788544314e-05, + "loss": 0.0191, + "step": 3310 + }, + { + "epoch": 5.6752136752136755, + "grad_norm": 0.10629488527774811, + "learning_rate": 2.5982514056086238e-05, + "loss": 0.0182, + "step": 3320 + }, + { + "epoch": 5.6923076923076925, + "grad_norm": 0.11531593650579453, + "learning_rate": 2.5954316918733992e-05, + "loss": 0.0187, + "step": 3330 + }, + { + "epoch": 5.7094017094017095, + "grad_norm": 0.11777219921350479, + "learning_rate": 2.5926036590626093e-05, + "loss": 0.0192, + "step": 3340 + }, + { + "epoch": 5.726495726495727, + "grad_norm": 0.11960875242948532, + "learning_rate": 2.589767328653283e-05, + "loss": 0.0183, + "step": 3350 + }, + { + "epoch": 5.743589743589744, + "grad_norm": 0.08489462733268738, + "learning_rate": 2.5869227221854646e-05, + "loss": 0.0179, + "step": 3360 + }, + { + "epoch": 5.760683760683761, + "grad_norm": 0.07769697159528732, + "learning_rate": 2.584069861262049e-05, + "loss": 0.0186, + "step": 3370 + }, + { + "epoch": 5.777777777777778, + "grad_norm": 0.10025164484977722, + "learning_rate": 2.58120876754862e-05, + "loss": 0.0143, + "step": 3380 + }, + { + "epoch": 5.794871794871795, + "grad_norm": 0.09745113551616669, + "learning_rate": 2.5783394627732815e-05, + "loss": 0.0164, + "step": 3390 + }, + { + "epoch": 5.811965811965812, + "grad_norm": 0.09607358276844025, + "learning_rate": 2.5754619687264957e-05, + "loss": 0.0205, + "step": 3400 + }, + { + "epoch": 5.829059829059829, + "grad_norm": 0.0831088125705719, + "learning_rate": 2.5725763072609178e-05, + "loss": 0.0166, + "step": 3410 + }, + { + "epoch": 5.846153846153846, + "grad_norm": 0.07195224612951279, + "learning_rate": 2.569682500291229e-05, + "loss": 0.0152, + "step": 3420 + }, + { + "epoch": 5.863247863247864, + "grad_norm": 0.09906742721796036, + "learning_rate": 2.5667805697939687e-05, + "loss": 0.0162, + "step": 3430 + }, + { + "epoch": 5.880341880341881, + "grad_norm": 0.07969392836093903, + "learning_rate": 2.5638705378073704e-05, + "loss": 0.0193, + "step": 3440 + }, + { + "epoch": 5.897435897435898, + "grad_norm": 0.0873081237077713, + "learning_rate": 2.5609524264311934e-05, + "loss": 0.0177, + "step": 3450 + }, + { + "epoch": 5.914529914529915, + "grad_norm": 0.0908493623137474, + "learning_rate": 2.5580262578265536e-05, + "loss": 0.0205, + "step": 3460 + }, + { + "epoch": 5.931623931623932, + "grad_norm": 0.08355225622653961, + "learning_rate": 2.5550920542157557e-05, + "loss": 0.0161, + "step": 3470 + }, + { + "epoch": 5.948717948717949, + "grad_norm": 0.06161191314458847, + "learning_rate": 2.5521498378821273e-05, + "loss": 0.0183, + "step": 3480 + }, + { + "epoch": 5.965811965811966, + "grad_norm": 0.09668168425559998, + "learning_rate": 2.549199631169844e-05, + "loss": 0.0175, + "step": 3490 + }, + { + "epoch": 5.982905982905983, + "grad_norm": 0.07232195138931274, + "learning_rate": 2.5462414564837663e-05, + "loss": 0.0186, + "step": 3500 + }, + { + "epoch": 6.0, + "grad_norm": 0.12467692792415619, + "learning_rate": 2.543275336289264e-05, + "loss": 0.017, + "step": 3510 + }, + { + "epoch": 6.017094017094017, + "grad_norm": 0.11001081019639969, + "learning_rate": 2.540301293112048e-05, + "loss": 0.0168, + "step": 3520 + }, + { + "epoch": 6.034188034188034, + "grad_norm": 0.08639950305223465, + "learning_rate": 2.537319349538e-05, + "loss": 0.0193, + "step": 3530 + }, + { + "epoch": 6.051282051282051, + "grad_norm": 0.09567685425281525, + "learning_rate": 2.534329528212999e-05, + "loss": 0.0168, + "step": 3540 + }, + { + "epoch": 6.068376068376068, + "grad_norm": 0.08502615988254547, + "learning_rate": 2.531331851842751e-05, + "loss": 0.0174, + "step": 3550 + }, + { + "epoch": 6.085470085470085, + "grad_norm": 0.07735824584960938, + "learning_rate": 2.528326343192615e-05, + "loss": 0.0172, + "step": 3560 + }, + { + "epoch": 6.102564102564102, + "grad_norm": 0.06508814543485641, + "learning_rate": 2.5253130250874316e-05, + "loss": 0.0169, + "step": 3570 + }, + { + "epoch": 6.119658119658119, + "grad_norm": 0.11148368567228317, + "learning_rate": 2.5222919204113494e-05, + "loss": 0.0188, + "step": 3580 + }, + { + "epoch": 6.136752136752137, + "grad_norm": 0.11212950944900513, + "learning_rate": 2.5192630521076494e-05, + "loss": 0.0176, + "step": 3590 + }, + { + "epoch": 6.153846153846154, + "grad_norm": 0.09221594035625458, + "learning_rate": 2.5162264431785734e-05, + "loss": 0.0201, + "step": 3600 + }, + { + "epoch": 6.170940170940171, + "grad_norm": 0.10451870411634445, + "learning_rate": 2.5131821166851484e-05, + "loss": 0.0165, + "step": 3610 + }, + { + "epoch": 6.188034188034188, + "grad_norm": 0.09675891697406769, + "learning_rate": 2.5101300957470095e-05, + "loss": 0.0165, + "step": 3620 + }, + { + "epoch": 6.205128205128205, + "grad_norm": 0.07100088149309158, + "learning_rate": 2.507070403542228e-05, + "loss": 0.0156, + "step": 3630 + }, + { + "epoch": 6.222222222222222, + "grad_norm": 0.09254378825426102, + "learning_rate": 2.504003063307132e-05, + "loss": 0.0171, + "step": 3640 + }, + { + "epoch": 6.239316239316239, + "grad_norm": 0.07285033911466599, + "learning_rate": 2.500928098336131e-05, + "loss": 0.017, + "step": 3650 + }, + { + "epoch": 6.256410256410256, + "grad_norm": 0.10864458978176117, + "learning_rate": 2.497845531981541e-05, + "loss": 0.0171, + "step": 3660 + }, + { + "epoch": 6.273504273504273, + "grad_norm": 0.09306647628545761, + "learning_rate": 2.4947553876534047e-05, + "loss": 0.0173, + "step": 3670 + }, + { + "epoch": 6.2905982905982905, + "grad_norm": 0.08375543355941772, + "learning_rate": 2.4916576888193137e-05, + "loss": 0.0158, + "step": 3680 + }, + { + "epoch": 6.3076923076923075, + "grad_norm": 0.10552448034286499, + "learning_rate": 2.488552459004231e-05, + "loss": 0.0174, + "step": 3690 + }, + { + "epoch": 6.3247863247863245, + "grad_norm": 0.07783381640911102, + "learning_rate": 2.4854397217903145e-05, + "loss": 0.0177, + "step": 3700 + }, + { + "epoch": 6.3418803418803416, + "grad_norm": 0.10520699620246887, + "learning_rate": 2.4823195008167325e-05, + "loss": 0.0173, + "step": 3710 + }, + { + "epoch": 6.358974358974359, + "grad_norm": 0.07988301664590836, + "learning_rate": 2.4791918197794907e-05, + "loss": 0.0173, + "step": 3720 + }, + { + "epoch": 6.3760683760683765, + "grad_norm": 0.09742307662963867, + "learning_rate": 2.4760567024312476e-05, + "loss": 0.0166, + "step": 3730 + }, + { + "epoch": 6.3931623931623935, + "grad_norm": 0.08026914298534393, + "learning_rate": 2.4729141725811344e-05, + "loss": 0.0183, + "step": 3740 + }, + { + "epoch": 6.410256410256411, + "grad_norm": 0.11827661842107773, + "learning_rate": 2.4697642540945774e-05, + "loss": 0.0182, + "step": 3750 + }, + { + "epoch": 6.427350427350428, + "grad_norm": 0.09172654151916504, + "learning_rate": 2.4666069708931136e-05, + "loss": 0.0175, + "step": 3760 + }, + { + "epoch": 6.444444444444445, + "grad_norm": 0.08934371173381805, + "learning_rate": 2.46344234695421e-05, + "loss": 0.0178, + "step": 3770 + }, + { + "epoch": 6.461538461538462, + "grad_norm": 0.0789095088839531, + "learning_rate": 2.460270406311082e-05, + "loss": 0.0165, + "step": 3780 + }, + { + "epoch": 6.478632478632479, + "grad_norm": 0.11394141614437103, + "learning_rate": 2.457091173052511e-05, + "loss": 0.0187, + "step": 3790 + }, + { + "epoch": 6.495726495726496, + "grad_norm": 0.08394981920719147, + "learning_rate": 2.4539046713226593e-05, + "loss": 0.0169, + "step": 3800 + }, + { + "epoch": 6.512820512820513, + "grad_norm": 0.08010982722043991, + "learning_rate": 2.4507109253208898e-05, + "loss": 0.0145, + "step": 3810 + }, + { + "epoch": 6.52991452991453, + "grad_norm": 0.09592076390981674, + "learning_rate": 2.4475099593015813e-05, + "loss": 0.0159, + "step": 3820 + }, + { + "epoch": 6.547008547008547, + "grad_norm": 0.0980319082736969, + "learning_rate": 2.4443017975739423e-05, + "loss": 0.018, + "step": 3830 + }, + { + "epoch": 6.564102564102564, + "grad_norm": 0.09010566025972366, + "learning_rate": 2.4410864645018288e-05, + "loss": 0.0159, + "step": 3840 + }, + { + "epoch": 6.581196581196581, + "grad_norm": 0.10003632307052612, + "learning_rate": 2.4378639845035583e-05, + "loss": 0.0168, + "step": 3850 + }, + { + "epoch": 6.598290598290598, + "grad_norm": 0.09281107783317566, + "learning_rate": 2.4346343820517237e-05, + "loss": 0.0158, + "step": 3860 + }, + { + "epoch": 6.615384615384615, + "grad_norm": 0.10115903615951538, + "learning_rate": 2.4313976816730095e-05, + "loss": 0.0175, + "step": 3870 + }, + { + "epoch": 6.632478632478632, + "grad_norm": 0.07335299998521805, + "learning_rate": 2.4281539079480047e-05, + "loss": 0.0145, + "step": 3880 + }, + { + "epoch": 6.64957264957265, + "grad_norm": 0.09117656201124191, + "learning_rate": 2.4249030855110128e-05, + "loss": 0.0157, + "step": 3890 + }, + { + "epoch": 6.666666666666667, + "grad_norm": 0.0800071731209755, + "learning_rate": 2.42164523904987e-05, + "loss": 0.0165, + "step": 3900 + }, + { + "epoch": 6.683760683760684, + "grad_norm": 0.0696878433227539, + "learning_rate": 2.4183803933057554e-05, + "loss": 0.0174, + "step": 3910 + }, + { + "epoch": 6.700854700854701, + "grad_norm": 0.0962003841996193, + "learning_rate": 2.4151085730730013e-05, + "loss": 0.0182, + "step": 3920 + }, + { + "epoch": 6.717948717948718, + "grad_norm": 0.0997065007686615, + "learning_rate": 2.4118298031989084e-05, + "loss": 0.0175, + "step": 3930 + }, + { + "epoch": 6.735042735042735, + "grad_norm": 0.10266530513763428, + "learning_rate": 2.4085441085835534e-05, + "loss": 0.0164, + "step": 3940 + }, + { + "epoch": 6.752136752136752, + "grad_norm": 0.09127513319253922, + "learning_rate": 2.405251514179604e-05, + "loss": 0.0175, + "step": 3950 + }, + { + "epoch": 6.769230769230769, + "grad_norm": 0.06895238906145096, + "learning_rate": 2.401952044992126e-05, + "loss": 0.017, + "step": 3960 + }, + { + "epoch": 6.786324786324786, + "grad_norm": 0.08264075219631195, + "learning_rate": 2.3986457260783942e-05, + "loss": 0.0175, + "step": 3970 + }, + { + "epoch": 6.803418803418803, + "grad_norm": 0.13420195877552032, + "learning_rate": 2.395332582547703e-05, + "loss": 0.0209, + "step": 3980 + }, + { + "epoch": 6.82051282051282, + "grad_norm": 0.10509937256574631, + "learning_rate": 2.3920126395611755e-05, + "loss": 0.0182, + "step": 3990 + }, + { + "epoch": 6.837606837606837, + "grad_norm": 0.08453608304262161, + "learning_rate": 2.3886859223315725e-05, + "loss": 0.0183, + "step": 4000 + }, + { + "epoch": 6.854700854700854, + "grad_norm": 0.07713332772254944, + "learning_rate": 2.385352456123099e-05, + "loss": 0.0168, + "step": 4010 + }, + { + "epoch": 6.871794871794872, + "grad_norm": 0.08435060828924179, + "learning_rate": 2.3820122662512155e-05, + "loss": 0.017, + "step": 4020 + }, + { + "epoch": 6.888888888888889, + "grad_norm": 0.07585082948207855, + "learning_rate": 2.3786653780824443e-05, + "loss": 0.0158, + "step": 4030 + }, + { + "epoch": 6.905982905982906, + "grad_norm": 0.09102822840213776, + "learning_rate": 2.3753118170341762e-05, + "loss": 0.0163, + "step": 4040 + }, + { + "epoch": 6.923076923076923, + "grad_norm": 0.08681073039770126, + "learning_rate": 2.3719516085744782e-05, + "loss": 0.0149, + "step": 4050 + }, + { + "epoch": 6.94017094017094, + "grad_norm": 0.09147541224956512, + "learning_rate": 2.3685847782218997e-05, + "loss": 0.0153, + "step": 4060 + }, + { + "epoch": 6.957264957264957, + "grad_norm": 0.11664493381977081, + "learning_rate": 2.3652113515452792e-05, + "loss": 0.0169, + "step": 4070 + }, + { + "epoch": 6.9743589743589745, + "grad_norm": 0.08450709283351898, + "learning_rate": 2.3618313541635506e-05, + "loss": 0.0172, + "step": 4080 + }, + { + "epoch": 6.9914529914529915, + "grad_norm": 0.09218313544988632, + "learning_rate": 2.358444811745547e-05, + "loss": 0.0145, + "step": 4090 + }, + { + "epoch": 7.0085470085470085, + "grad_norm": 0.1001223474740982, + "learning_rate": 2.3550517500098068e-05, + "loss": 0.0158, + "step": 4100 + }, + { + "epoch": 7.0256410256410255, + "grad_norm": 0.07710881531238556, + "learning_rate": 2.3516521947243787e-05, + "loss": 0.016, + "step": 4110 + }, + { + "epoch": 7.042735042735043, + "grad_norm": 0.0710505023598671, + "learning_rate": 2.348246171706626e-05, + "loss": 0.0151, + "step": 4120 + }, + { + "epoch": 7.05982905982906, + "grad_norm": 0.08820394426584244, + "learning_rate": 2.3448337068230288e-05, + "loss": 0.0154, + "step": 4130 + }, + { + "epoch": 7.076923076923077, + "grad_norm": 0.09810654073953629, + "learning_rate": 2.3414148259889896e-05, + "loss": 0.0179, + "step": 4140 + }, + { + "epoch": 7.094017094017094, + "grad_norm": 0.07964249700307846, + "learning_rate": 2.3379895551686358e-05, + "loss": 0.0166, + "step": 4150 + }, + { + "epoch": 7.111111111111111, + "grad_norm": 0.10602698475122452, + "learning_rate": 2.334557920374623e-05, + "loss": 0.0179, + "step": 4160 + }, + { + "epoch": 7.128205128205128, + "grad_norm": 0.06306713074445724, + "learning_rate": 2.3311199476679367e-05, + "loss": 0.0159, + "step": 4170 + }, + { + "epoch": 7.145299145299146, + "grad_norm": 0.09402914345264435, + "learning_rate": 2.327675663157693e-05, + "loss": 0.0164, + "step": 4180 + }, + { + "epoch": 7.162393162393163, + "grad_norm": 0.10625981539487839, + "learning_rate": 2.3242250930009446e-05, + "loss": 0.0157, + "step": 4190 + }, + { + "epoch": 7.17948717948718, + "grad_norm": 0.07685832679271698, + "learning_rate": 2.3207682634024777e-05, + "loss": 0.016, + "step": 4200 + }, + { + "epoch": 7.196581196581197, + "grad_norm": 0.07576439529657364, + "learning_rate": 2.3173052006146155e-05, + "loss": 0.0145, + "step": 4210 + }, + { + "epoch": 7.213675213675214, + "grad_norm": 0.0860842764377594, + "learning_rate": 2.313835930937018e-05, + "loss": 0.0152, + "step": 4220 + }, + { + "epoch": 7.230769230769231, + "grad_norm": 0.06694341450929642, + "learning_rate": 2.310360480716483e-05, + "loss": 0.0139, + "step": 4230 + }, + { + "epoch": 7.247863247863248, + "grad_norm": 0.07913471013307571, + "learning_rate": 2.3068788763467434e-05, + "loss": 0.0138, + "step": 4240 + }, + { + "epoch": 7.264957264957265, + "grad_norm": 0.0752611756324768, + "learning_rate": 2.3033911442682708e-05, + "loss": 0.0173, + "step": 4250 + }, + { + "epoch": 7.282051282051282, + "grad_norm": 0.09453887492418289, + "learning_rate": 2.299897310968072e-05, + "loss": 0.0161, + "step": 4260 + }, + { + "epoch": 7.299145299145299, + "grad_norm": 0.07289311289787292, + "learning_rate": 2.2963974029794893e-05, + "loss": 0.014, + "step": 4270 + }, + { + "epoch": 7.316239316239316, + "grad_norm": 0.07161838561296463, + "learning_rate": 2.2928914468819968e-05, + "loss": 0.0172, + "step": 4280 + }, + { + "epoch": 7.333333333333333, + "grad_norm": 0.09718061983585358, + "learning_rate": 2.2893794693010002e-05, + "loss": 0.0171, + "step": 4290 + }, + { + "epoch": 7.35042735042735, + "grad_norm": 0.08493733406066895, + "learning_rate": 2.2858614969076358e-05, + "loss": 0.0156, + "step": 4300 + }, + { + "epoch": 7.367521367521368, + "grad_norm": 0.09210729598999023, + "learning_rate": 2.2823375564185645e-05, + "loss": 0.0177, + "step": 4310 + }, + { + "epoch": 7.384615384615385, + "grad_norm": 0.08660814166069031, + "learning_rate": 2.278807674595773e-05, + "loss": 0.016, + "step": 4320 + }, + { + "epoch": 7.401709401709402, + "grad_norm": 0.0879073292016983, + "learning_rate": 2.275271878246367e-05, + "loss": 0.0155, + "step": 4330 + }, + { + "epoch": 7.418803418803419, + "grad_norm": 0.0978168323636055, + "learning_rate": 2.271730194222369e-05, + "loss": 0.0167, + "step": 4340 + }, + { + "epoch": 7.435897435897436, + "grad_norm": 0.10353145748376846, + "learning_rate": 2.2681826494205156e-05, + "loss": 0.0168, + "step": 4350 + }, + { + "epoch": 7.452991452991453, + "grad_norm": 0.08525838702917099, + "learning_rate": 2.2646292707820513e-05, + "loss": 0.0155, + "step": 4360 + }, + { + "epoch": 7.47008547008547, + "grad_norm": 0.09657790511846542, + "learning_rate": 2.2610700852925254e-05, + "loss": 0.0153, + "step": 4370 + }, + { + "epoch": 7.487179487179487, + "grad_norm": 0.10523585230112076, + "learning_rate": 2.2575051199815854e-05, + "loss": 0.0172, + "step": 4380 + }, + { + "epoch": 7.504273504273504, + "grad_norm": 0.07577922940254211, + "learning_rate": 2.253934401922773e-05, + "loss": 0.0167, + "step": 4390 + }, + { + "epoch": 7.521367521367521, + "grad_norm": 0.09458927065134048, + "learning_rate": 2.250357958233319e-05, + "loss": 0.0178, + "step": 4400 + }, + { + "epoch": 7.538461538461538, + "grad_norm": 0.08498671650886536, + "learning_rate": 2.2467758160739352e-05, + "loss": 0.0165, + "step": 4410 + }, + { + "epoch": 7.555555555555555, + "grad_norm": 0.07391520589590073, + "learning_rate": 2.2431880026486114e-05, + "loss": 0.0163, + "step": 4420 + }, + { + "epoch": 7.572649572649572, + "grad_norm": 0.08839573711156845, + "learning_rate": 2.2395945452044045e-05, + "loss": 0.0187, + "step": 4430 + }, + { + "epoch": 7.589743589743589, + "grad_norm": 0.09913501143455505, + "learning_rate": 2.2359954710312363e-05, + "loss": 0.0141, + "step": 4440 + }, + { + "epoch": 7.6068376068376065, + "grad_norm": 0.12251193076372147, + "learning_rate": 2.232390807461682e-05, + "loss": 0.016, + "step": 4450 + }, + { + "epoch": 7.6239316239316235, + "grad_norm": 0.10324403643608093, + "learning_rate": 2.2287805818707658e-05, + "loss": 0.0174, + "step": 4460 + }, + { + "epoch": 7.641025641025641, + "grad_norm": 0.08906015753746033, + "learning_rate": 2.2251648216757514e-05, + "loss": 0.0147, + "step": 4470 + }, + { + "epoch": 7.6581196581196584, + "grad_norm": 0.09674681723117828, + "learning_rate": 2.2215435543359337e-05, + "loss": 0.0149, + "step": 4480 + }, + { + "epoch": 7.6752136752136755, + "grad_norm": 0.0863393247127533, + "learning_rate": 2.2179168073524308e-05, + "loss": 0.0164, + "step": 4490 + }, + { + "epoch": 7.6923076923076925, + "grad_norm": 0.07127152383327484, + "learning_rate": 2.2142846082679763e-05, + "loss": 0.0164, + "step": 4500 + }, + { + "epoch": 7.7094017094017095, + "grad_norm": 0.07405145466327667, + "learning_rate": 2.2106469846667078e-05, + "loss": 0.0147, + "step": 4510 + }, + { + "epoch": 7.726495726495727, + "grad_norm": 0.0741042047739029, + "learning_rate": 2.2070039641739578e-05, + "loss": 0.0168, + "step": 4520 + }, + { + "epoch": 7.743589743589744, + "grad_norm": 0.09215082973241806, + "learning_rate": 2.2033555744560462e-05, + "loss": 0.0156, + "step": 4530 + }, + { + "epoch": 7.760683760683761, + "grad_norm": 0.06406484544277191, + "learning_rate": 2.1997018432200674e-05, + "loss": 0.0153, + "step": 4540 + }, + { + "epoch": 7.777777777777778, + "grad_norm": 0.0733003243803978, + "learning_rate": 2.1960427982136818e-05, + "loss": 0.0143, + "step": 4550 + }, + { + "epoch": 7.794871794871795, + "grad_norm": 0.07966963946819305, + "learning_rate": 2.1923784672249038e-05, + "loss": 0.0168, + "step": 4560 + }, + { + "epoch": 7.811965811965812, + "grad_norm": 0.08074156194925308, + "learning_rate": 2.1887088780818918e-05, + "loss": 0.014, + "step": 4570 + }, + { + "epoch": 7.829059829059829, + "grad_norm": 0.10418794304132462, + "learning_rate": 2.1850340586527354e-05, + "loss": 0.0171, + "step": 4580 + }, + { + "epoch": 7.846153846153846, + "grad_norm": 0.0782044306397438, + "learning_rate": 2.1813540368452468e-05, + "loss": 0.0151, + "step": 4590 + }, + { + "epoch": 7.863247863247864, + "grad_norm": 0.08316248655319214, + "learning_rate": 2.1776688406067447e-05, + "loss": 0.0157, + "step": 4600 + }, + { + "epoch": 7.880341880341881, + "grad_norm": 0.08307249844074249, + "learning_rate": 2.1739784979238447e-05, + "loss": 0.0154, + "step": 4610 + }, + { + "epoch": 7.897435897435898, + "grad_norm": 0.05799156054854393, + "learning_rate": 2.1702830368222478e-05, + "loss": 0.0165, + "step": 4620 + }, + { + "epoch": 7.914529914529915, + "grad_norm": 0.0774935930967331, + "learning_rate": 2.166582485366523e-05, + "loss": 0.0156, + "step": 4630 + }, + { + "epoch": 7.931623931623932, + "grad_norm": 0.07723759859800339, + "learning_rate": 2.1628768716599e-05, + "loss": 0.0147, + "step": 4640 + }, + { + "epoch": 7.948717948717949, + "grad_norm": 0.09672164171934128, + "learning_rate": 2.1591662238440515e-05, + "loss": 0.0155, + "step": 4650 + }, + { + "epoch": 7.965811965811966, + "grad_norm": 0.0799407809972763, + "learning_rate": 2.1554505700988813e-05, + "loss": 0.0139, + "step": 4660 + }, + { + "epoch": 7.982905982905983, + "grad_norm": 0.0932922214269638, + "learning_rate": 2.1517299386423094e-05, + "loss": 0.0158, + "step": 4670 + }, + { + "epoch": 8.0, + "grad_norm": 0.0904204398393631, + "learning_rate": 2.148004357730059e-05, + "loss": 0.0155, + "step": 4680 + }, + { + "epoch": 8.017094017094017, + "grad_norm": 0.08726400136947632, + "learning_rate": 2.1442738556554404e-05, + "loss": 0.0165, + "step": 4690 + }, + { + "epoch": 8.034188034188034, + "grad_norm": 0.08845978230237961, + "learning_rate": 2.1405384607491375e-05, + "loss": 0.0175, + "step": 4700 + }, + { + "epoch": 8.051282051282051, + "grad_norm": 0.08011137694120407, + "learning_rate": 2.1367982013789913e-05, + "loss": 0.0155, + "step": 4710 + }, + { + "epoch": 8.068376068376068, + "grad_norm": 0.11915645003318787, + "learning_rate": 2.133053105949786e-05, + "loss": 0.0154, + "step": 4720 + }, + { + "epoch": 8.085470085470085, + "grad_norm": 0.06125086173415184, + "learning_rate": 2.129303202903032e-05, + "loss": 0.0162, + "step": 4730 + }, + { + "epoch": 8.102564102564102, + "grad_norm": 0.07610971480607986, + "learning_rate": 2.1255485207167504e-05, + "loss": 0.0156, + "step": 4740 + }, + { + "epoch": 8.11965811965812, + "grad_norm": 0.11904177814722061, + "learning_rate": 2.1217890879052556e-05, + "loss": 0.0154, + "step": 4750 + }, + { + "epoch": 8.136752136752136, + "grad_norm": 0.06247622147202492, + "learning_rate": 2.1180249330189423e-05, + "loss": 0.0167, + "step": 4760 + }, + { + "epoch": 8.153846153846153, + "grad_norm": 0.0804535299539566, + "learning_rate": 2.1142560846440645e-05, + "loss": 0.0171, + "step": 4770 + }, + { + "epoch": 8.17094017094017, + "grad_norm": 0.08552727848291397, + "learning_rate": 2.1104825714025203e-05, + "loss": 0.016, + "step": 4780 + }, + { + "epoch": 8.188034188034187, + "grad_norm": 0.06268288195133209, + "learning_rate": 2.106704421951634e-05, + "loss": 0.0165, + "step": 4790 + }, + { + "epoch": 8.205128205128204, + "grad_norm": 0.0662277489900589, + "learning_rate": 2.1029216649839404e-05, + "loss": 0.0138, + "step": 4800 + }, + { + "epoch": 8.222222222222221, + "grad_norm": 0.08757280558347702, + "learning_rate": 2.099134329226963e-05, + "loss": 0.0162, + "step": 4810 + }, + { + "epoch": 8.239316239316238, + "grad_norm": 0.08624053746461868, + "learning_rate": 2.0953424434430006e-05, + "loss": 0.0154, + "step": 4820 + }, + { + "epoch": 8.256410256410255, + "grad_norm": 0.10608145594596863, + "learning_rate": 2.091546036428905e-05, + "loss": 0.0151, + "step": 4830 + }, + { + "epoch": 8.273504273504274, + "grad_norm": 0.06986302137374878, + "learning_rate": 2.0877451370158633e-05, + "loss": 0.0157, + "step": 4840 + }, + { + "epoch": 8.290598290598291, + "grad_norm": 0.08707921206951141, + "learning_rate": 2.083939774069181e-05, + "loss": 0.0141, + "step": 4850 + }, + { + "epoch": 8.307692307692308, + "grad_norm": 0.07429157197475433, + "learning_rate": 2.080129976488059e-05, + "loss": 0.0145, + "step": 4860 + }, + { + "epoch": 8.324786324786325, + "grad_norm": 0.0858520120382309, + "learning_rate": 2.0763157732053795e-05, + "loss": 0.0155, + "step": 4870 + }, + { + "epoch": 8.341880341880342, + "grad_norm": 0.096824049949646, + "learning_rate": 2.0724971931874796e-05, + "loss": 0.0148, + "step": 4880 + }, + { + "epoch": 8.35897435897436, + "grad_norm": 0.061589065939188004, + "learning_rate": 2.0686742654339363e-05, + "loss": 0.0148, + "step": 4890 + }, + { + "epoch": 8.376068376068377, + "grad_norm": 0.08161415159702301, + "learning_rate": 2.0648470189773455e-05, + "loss": 0.0142, + "step": 4900 + }, + { + "epoch": 8.393162393162394, + "grad_norm": 0.057990241795778275, + "learning_rate": 2.0610154828830997e-05, + "loss": 0.0139, + "step": 4910 + }, + { + "epoch": 8.41025641025641, + "grad_norm": 0.09365546703338623, + "learning_rate": 2.0571796862491686e-05, + "loss": 0.0167, + "step": 4920 + }, + { + "epoch": 8.427350427350428, + "grad_norm": 0.11589137464761734, + "learning_rate": 2.0533396582058783e-05, + "loss": 0.0169, + "step": 4930 + }, + { + "epoch": 8.444444444444445, + "grad_norm": 0.07843475043773651, + "learning_rate": 2.0494954279156893e-05, + "loss": 0.0138, + "step": 4940 + }, + { + "epoch": 8.461538461538462, + "grad_norm": 0.07571505755186081, + "learning_rate": 2.0456470245729748e-05, + "loss": 0.016, + "step": 4950 + }, + { + "epoch": 8.478632478632479, + "grad_norm": 0.09430881589651108, + "learning_rate": 2.0417944774038013e-05, + "loss": 0.0165, + "step": 4960 + }, + { + "epoch": 8.495726495726496, + "grad_norm": 0.08276867121458054, + "learning_rate": 2.0379378156657026e-05, + "loss": 0.0146, + "step": 4970 + }, + { + "epoch": 8.512820512820513, + "grad_norm": 0.07636648416519165, + "learning_rate": 2.0340770686474618e-05, + "loss": 0.0149, + "step": 4980 + }, + { + "epoch": 8.52991452991453, + "grad_norm": 0.08420944213867188, + "learning_rate": 2.030212265668886e-05, + "loss": 0.0137, + "step": 4990 + }, + { + "epoch": 8.547008547008547, + "grad_norm": 0.0849870890378952, + "learning_rate": 2.0263434360805847e-05, + "loss": 0.0169, + "step": 5000 + }, + { + "epoch": 8.564102564102564, + "grad_norm": 0.08794967085123062, + "learning_rate": 2.0224706092637472e-05, + "loss": 0.0149, + "step": 5010 + }, + { + "epoch": 8.581196581196581, + "grad_norm": 0.10814186930656433, + "learning_rate": 2.018593814629919e-05, + "loss": 0.0158, + "step": 5020 + }, + { + "epoch": 8.598290598290598, + "grad_norm": 0.08679825812578201, + "learning_rate": 2.0147130816207786e-05, + "loss": 0.0155, + "step": 5030 + }, + { + "epoch": 8.615384615384615, + "grad_norm": 0.08614245802164078, + "learning_rate": 2.0108284397079127e-05, + "loss": 0.0133, + "step": 5040 + }, + { + "epoch": 8.632478632478632, + "grad_norm": 0.08051575720310211, + "learning_rate": 2.006939918392596e-05, + "loss": 0.0137, + "step": 5050 + }, + { + "epoch": 8.649572649572649, + "grad_norm": 0.14550358057022095, + "learning_rate": 2.0030475472055625e-05, + "loss": 0.0156, + "step": 5060 + }, + { + "epoch": 8.666666666666666, + "grad_norm": 0.07240953296422958, + "learning_rate": 1.9991513557067835e-05, + "loss": 0.0136, + "step": 5070 + }, + { + "epoch": 8.683760683760683, + "grad_norm": 0.09340184181928635, + "learning_rate": 1.9952513734852444e-05, + "loss": 0.0139, + "step": 5080 + }, + { + "epoch": 8.7008547008547, + "grad_norm": 0.08796098083257675, + "learning_rate": 1.991347630158717e-05, + "loss": 0.0145, + "step": 5090 + }, + { + "epoch": 8.717948717948717, + "grad_norm": 0.07233105599880219, + "learning_rate": 1.9874401553735385e-05, + "loss": 0.0156, + "step": 5100 + }, + { + "epoch": 8.735042735042736, + "grad_norm": 0.08621574193239212, + "learning_rate": 1.9835289788043814e-05, + "loss": 0.0157, + "step": 5110 + }, + { + "epoch": 8.752136752136753, + "grad_norm": 0.08131912350654602, + "learning_rate": 1.979614130154032e-05, + "loss": 0.0151, + "step": 5120 + }, + { + "epoch": 8.76923076923077, + "grad_norm": 0.0870567336678505, + "learning_rate": 1.9756956391531646e-05, + "loss": 0.0143, + "step": 5130 + }, + { + "epoch": 8.786324786324787, + "grad_norm": 0.07275652140378952, + "learning_rate": 1.9717735355601126e-05, + "loss": 0.0156, + "step": 5140 + }, + { + "epoch": 8.803418803418804, + "grad_norm": 0.07009653747081757, + "learning_rate": 1.9678478491606462e-05, + "loss": 0.0145, + "step": 5150 + }, + { + "epoch": 8.820512820512821, + "grad_norm": 0.09490709006786346, + "learning_rate": 1.9639186097677442e-05, + "loss": 0.0141, + "step": 5160 + }, + { + "epoch": 8.837606837606838, + "grad_norm": 0.07979147881269455, + "learning_rate": 1.9599858472213682e-05, + "loss": 0.0147, + "step": 5170 + }, + { + "epoch": 8.854700854700855, + "grad_norm": 0.08306172490119934, + "learning_rate": 1.9560495913882353e-05, + "loss": 0.013, + "step": 5180 + }, + { + "epoch": 8.871794871794872, + "grad_norm": 0.12175660580396652, + "learning_rate": 1.9521098721615923e-05, + "loss": 0.0146, + "step": 5190 + }, + { + "epoch": 8.88888888888889, + "grad_norm": 0.06418270617723465, + "learning_rate": 1.9481667194609882e-05, + "loss": 0.015, + "step": 5200 + }, + { + "epoch": 8.905982905982906, + "grad_norm": 0.07214277237653732, + "learning_rate": 1.9442201632320464e-05, + "loss": 0.0138, + "step": 5210 + }, + { + "epoch": 8.923076923076923, + "grad_norm": 0.06605114787817001, + "learning_rate": 1.9402702334462383e-05, + "loss": 0.015, + "step": 5220 + }, + { + "epoch": 8.94017094017094, + "grad_norm": 0.0511237196624279, + "learning_rate": 1.9363169601006552e-05, + "loss": 0.0146, + "step": 5230 + }, + { + "epoch": 8.957264957264957, + "grad_norm": 0.0878906175494194, + "learning_rate": 1.9323603732177798e-05, + "loss": 0.0172, + "step": 5240 + }, + { + "epoch": 8.974358974358974, + "grad_norm": 0.09568879753351212, + "learning_rate": 1.9284005028452607e-05, + "loss": 0.0142, + "step": 5250 + }, + { + "epoch": 8.991452991452991, + "grad_norm": 0.07162164151668549, + "learning_rate": 1.9244373790556803e-05, + "loss": 0.0154, + "step": 5260 + }, + { + "epoch": 9.008547008547009, + "grad_norm": 0.07112082093954086, + "learning_rate": 1.9204710319463297e-05, + "loss": 0.0146, + "step": 5270 + }, + { + "epoch": 9.025641025641026, + "grad_norm": 0.07526064664125443, + "learning_rate": 1.9165014916389792e-05, + "loss": 0.0147, + "step": 5280 + }, + { + "epoch": 9.042735042735043, + "grad_norm": 0.06581240892410278, + "learning_rate": 1.9125287882796483e-05, + "loss": 0.0142, + "step": 5290 + }, + { + "epoch": 9.05982905982906, + "grad_norm": 0.07891480624675751, + "learning_rate": 1.9085529520383783e-05, + "loss": 0.0138, + "step": 5300 + }, + { + "epoch": 9.076923076923077, + "grad_norm": 0.07262127101421356, + "learning_rate": 1.9045740131090037e-05, + "loss": 0.014, + "step": 5310 + }, + { + "epoch": 9.094017094017094, + "grad_norm": 0.09044957906007767, + "learning_rate": 1.9005920017089196e-05, + "loss": 0.0146, + "step": 5320 + }, + { + "epoch": 9.11111111111111, + "grad_norm": 0.06296322494745255, + "learning_rate": 1.8966069480788562e-05, + "loss": 0.0135, + "step": 5330 + }, + { + "epoch": 9.128205128205128, + "grad_norm": 0.05622626468539238, + "learning_rate": 1.8926188824826477e-05, + "loss": 0.0141, + "step": 5340 + }, + { + "epoch": 9.145299145299145, + "grad_norm": 0.09232998639345169, + "learning_rate": 1.8886278352070002e-05, + "loss": 0.0159, + "step": 5350 + }, + { + "epoch": 9.162393162393162, + "grad_norm": 0.06181327998638153, + "learning_rate": 1.884633836561265e-05, + "loss": 0.0143, + "step": 5360 + }, + { + "epoch": 9.179487179487179, + "grad_norm": 0.0727447122335434, + "learning_rate": 1.8806369168772086e-05, + "loss": 0.0127, + "step": 5370 + }, + { + "epoch": 9.196581196581196, + "grad_norm": 0.06865410506725311, + "learning_rate": 1.8766371065087778e-05, + "loss": 0.0142, + "step": 5380 + }, + { + "epoch": 9.213675213675213, + "grad_norm": 0.07475120574235916, + "learning_rate": 1.8726344358318736e-05, + "loss": 0.0158, + "step": 5390 + }, + { + "epoch": 9.23076923076923, + "grad_norm": 0.066522017121315, + "learning_rate": 1.86862893524412e-05, + "loss": 0.014, + "step": 5400 + }, + { + "epoch": 9.247863247863247, + "grad_norm": 0.0740792527794838, + "learning_rate": 1.864620635164631e-05, + "loss": 0.0158, + "step": 5410 + }, + { + "epoch": 9.264957264957266, + "grad_norm": 0.07293172180652618, + "learning_rate": 1.8606095660337813e-05, + "loss": 0.0155, + "step": 5420 + }, + { + "epoch": 9.282051282051283, + "grad_norm": 0.05641787499189377, + "learning_rate": 1.8565957583129753e-05, + "loss": 0.0126, + "step": 5430 + }, + { + "epoch": 9.2991452991453, + "grad_norm": 0.09565582126379013, + "learning_rate": 1.852579242484414e-05, + "loss": 0.0139, + "step": 5440 + }, + { + "epoch": 9.316239316239317, + "grad_norm": 0.06965626031160355, + "learning_rate": 1.848560049050866e-05, + "loss": 0.0138, + "step": 5450 + }, + { + "epoch": 9.333333333333334, + "grad_norm": 0.0819479301571846, + "learning_rate": 1.8445382085354336e-05, + "loss": 0.0137, + "step": 5460 + }, + { + "epoch": 9.350427350427351, + "grad_norm": 0.07780192047357559, + "learning_rate": 1.8405137514813224e-05, + "loss": 0.0131, + "step": 5470 + }, + { + "epoch": 9.367521367521368, + "grad_norm": 0.07456802576780319, + "learning_rate": 1.8364867084516084e-05, + "loss": 0.0133, + "step": 5480 + }, + { + "epoch": 9.384615384615385, + "grad_norm": 0.07604533433914185, + "learning_rate": 1.8324571100290064e-05, + "loss": 0.015, + "step": 5490 + }, + { + "epoch": 9.401709401709402, + "grad_norm": 0.09989188611507416, + "learning_rate": 1.8284249868156378e-05, + "loss": 0.0146, + "step": 5500 + }, + { + "epoch": 9.418803418803419, + "grad_norm": 0.08827247470617294, + "learning_rate": 1.8243903694327992e-05, + "loss": 0.015, + "step": 5510 + }, + { + "epoch": 9.435897435897436, + "grad_norm": 0.06643637269735336, + "learning_rate": 1.8203532885207272e-05, + "loss": 0.0153, + "step": 5520 + }, + { + "epoch": 9.452991452991453, + "grad_norm": 0.05222713574767113, + "learning_rate": 1.8163137747383675e-05, + "loss": 0.0128, + "step": 5530 + }, + { + "epoch": 9.47008547008547, + "grad_norm": 0.0731734186410904, + "learning_rate": 1.812271858763143e-05, + "loss": 0.0153, + "step": 5540 + }, + { + "epoch": 9.487179487179487, + "grad_norm": 0.0945495218038559, + "learning_rate": 1.8082275712907187e-05, + "loss": 0.0136, + "step": 5550 + }, + { + "epoch": 9.504273504273504, + "grad_norm": 0.07506240904331207, + "learning_rate": 1.8041809430347688e-05, + "loss": 0.0165, + "step": 5560 + }, + { + "epoch": 9.521367521367521, + "grad_norm": 0.07505471259355545, + "learning_rate": 1.8001320047267465e-05, + "loss": 0.0137, + "step": 5570 + }, + { + "epoch": 9.538461538461538, + "grad_norm": 0.06896806508302689, + "learning_rate": 1.796080787115646e-05, + "loss": 0.0174, + "step": 5580 + }, + { + "epoch": 9.555555555555555, + "grad_norm": 0.08730415254831314, + "learning_rate": 1.7920273209677722e-05, + "loss": 0.0162, + "step": 5590 + }, + { + "epoch": 9.572649572649572, + "grad_norm": 0.06885059177875519, + "learning_rate": 1.7879716370665068e-05, + "loss": 0.0153, + "step": 5600 + }, + { + "epoch": 9.58974358974359, + "grad_norm": 0.07838878780603409, + "learning_rate": 1.783913766212073e-05, + "loss": 0.0152, + "step": 5610 + }, + { + "epoch": 9.606837606837606, + "grad_norm": 0.086198590695858, + "learning_rate": 1.779853739221302e-05, + "loss": 0.0158, + "step": 5620 + }, + { + "epoch": 9.623931623931623, + "grad_norm": 0.0976022407412529, + "learning_rate": 1.775791586927401e-05, + "loss": 0.0157, + "step": 5630 + }, + { + "epoch": 9.64102564102564, + "grad_norm": 0.0674310177564621, + "learning_rate": 1.771727340179716e-05, + "loss": 0.0123, + "step": 5640 + }, + { + "epoch": 9.658119658119658, + "grad_norm": 0.08769066631793976, + "learning_rate": 1.7676610298434993e-05, + "loss": 0.0144, + "step": 5650 + }, + { + "epoch": 9.675213675213675, + "grad_norm": 0.09793929755687714, + "learning_rate": 1.7635926867996758e-05, + "loss": 0.0141, + "step": 5660 + }, + { + "epoch": 9.692307692307692, + "grad_norm": 0.10376740992069244, + "learning_rate": 1.7595223419446064e-05, + "loss": 0.0145, + "step": 5670 + }, + { + "epoch": 9.709401709401709, + "grad_norm": 0.0791521668434143, + "learning_rate": 1.7554500261898545e-05, + "loss": 0.0137, + "step": 5680 + }, + { + "epoch": 9.726495726495726, + "grad_norm": 0.07923854142427444, + "learning_rate": 1.7513757704619513e-05, + "loss": 0.0146, + "step": 5690 + }, + { + "epoch": 9.743589743589745, + "grad_norm": 0.09366612136363983, + "learning_rate": 1.747299605702161e-05, + "loss": 0.0123, + "step": 5700 + }, + { + "epoch": 9.760683760683762, + "grad_norm": 0.0686226338148117, + "learning_rate": 1.7432215628662463e-05, + "loss": 0.0154, + "step": 5710 + }, + { + "epoch": 9.777777777777779, + "grad_norm": 0.0803999975323677, + "learning_rate": 1.7391416729242307e-05, + "loss": 0.0134, + "step": 5720 + }, + { + "epoch": 9.794871794871796, + "grad_norm": 0.08769650012254715, + "learning_rate": 1.7350599668601666e-05, + "loss": 0.0158, + "step": 5730 + }, + { + "epoch": 9.811965811965813, + "grad_norm": 0.05637761950492859, + "learning_rate": 1.730976475671899e-05, + "loss": 0.0165, + "step": 5740 + }, + { + "epoch": 9.82905982905983, + "grad_norm": 0.08623926341533661, + "learning_rate": 1.726891230370829e-05, + "loss": 0.0121, + "step": 5750 + }, + { + "epoch": 9.846153846153847, + "grad_norm": 0.07174214720726013, + "learning_rate": 1.7228042619816786e-05, + "loss": 0.0136, + "step": 5760 + }, + { + "epoch": 9.863247863247864, + "grad_norm": 0.07452727109193802, + "learning_rate": 1.7187156015422573e-05, + "loss": 0.0143, + "step": 5770 + }, + { + "epoch": 9.88034188034188, + "grad_norm": 0.09470614790916443, + "learning_rate": 1.7146252801032225e-05, + "loss": 0.0139, + "step": 5780 + }, + { + "epoch": 9.897435897435898, + "grad_norm": 0.07713735103607178, + "learning_rate": 1.7105333287278465e-05, + "loss": 0.0128, + "step": 5790 + }, + { + "epoch": 9.914529914529915, + "grad_norm": 0.06650368869304657, + "learning_rate": 1.706439778491781e-05, + "loss": 0.0137, + "step": 5800 + }, + { + "epoch": 9.931623931623932, + "grad_norm": 0.08439625799655914, + "learning_rate": 1.702344660482818e-05, + "loss": 0.0128, + "step": 5810 + }, + { + "epoch": 9.948717948717949, + "grad_norm": 0.09827966243028641, + "learning_rate": 1.6982480058006575e-05, + "loss": 0.0159, + "step": 5820 + }, + { + "epoch": 9.965811965811966, + "grad_norm": 0.0785556510090828, + "learning_rate": 1.6941498455566684e-05, + "loss": 0.0132, + "step": 5830 + }, + { + "epoch": 9.982905982905983, + "grad_norm": 0.08389803767204285, + "learning_rate": 1.690050210873654e-05, + "loss": 0.0129, + "step": 5840 + }, + { + "epoch": 10.0, + "grad_norm": 0.07077234238386154, + "learning_rate": 1.6859491328856143e-05, + "loss": 0.0141, + "step": 5850 + }, + { + "epoch": 10.017094017094017, + "grad_norm": 0.05990169942378998, + "learning_rate": 1.6818466427375114e-05, + "loss": 0.0137, + "step": 5860 + }, + { + "epoch": 10.034188034188034, + "grad_norm": 0.10044749826192856, + "learning_rate": 1.67774277158503e-05, + "loss": 0.0134, + "step": 5870 + }, + { + "epoch": 10.051282051282051, + "grad_norm": 0.08006548881530762, + "learning_rate": 1.673637550594344e-05, + "loss": 0.0134, + "step": 5880 + }, + { + "epoch": 10.068376068376068, + "grad_norm": 0.07858940958976746, + "learning_rate": 1.669531010941879e-05, + "loss": 0.0123, + "step": 5890 + }, + { + "epoch": 10.085470085470085, + "grad_norm": 0.08003196865320206, + "learning_rate": 1.6654231838140723e-05, + "loss": 0.0129, + "step": 5900 + }, + { + "epoch": 10.102564102564102, + "grad_norm": 0.06548362970352173, + "learning_rate": 1.661314100407142e-05, + "loss": 0.0131, + "step": 5910 + }, + { + "epoch": 10.11965811965812, + "grad_norm": 0.08702109009027481, + "learning_rate": 1.657203791926845e-05, + "loss": 0.0129, + "step": 5920 + }, + { + "epoch": 10.136752136752136, + "grad_norm": 0.10557379573583603, + "learning_rate": 1.653092289588241e-05, + "loss": 0.0136, + "step": 5930 + }, + { + "epoch": 10.153846153846153, + "grad_norm": 0.07397761195898056, + "learning_rate": 1.6489796246154586e-05, + "loss": 0.0155, + "step": 5940 + }, + { + "epoch": 10.17094017094017, + "grad_norm": 0.05981341376900673, + "learning_rate": 1.6448658282414543e-05, + "loss": 0.0132, + "step": 5950 + }, + { + "epoch": 10.188034188034187, + "grad_norm": 0.06545960903167725, + "learning_rate": 1.6407509317077764e-05, + "loss": 0.0145, + "step": 5960 + }, + { + "epoch": 10.205128205128204, + "grad_norm": 0.1132902055978775, + "learning_rate": 1.6366349662643297e-05, + "loss": 0.0135, + "step": 5970 + }, + { + "epoch": 10.222222222222221, + "grad_norm": 0.07090490311384201, + "learning_rate": 1.6325179631691355e-05, + "loss": 0.0126, + "step": 5980 + }, + { + "epoch": 10.239316239316238, + "grad_norm": 0.06559021025896072, + "learning_rate": 1.6283999536880957e-05, + "loss": 0.0157, + "step": 5990 + }, + { + "epoch": 10.256410256410255, + "grad_norm": 0.0638224333524704, + "learning_rate": 1.6242809690947555e-05, + "loss": 0.0114, + "step": 6000 + }, + { + "epoch": 10.273504273504274, + "grad_norm": 0.0664515271782875, + "learning_rate": 1.6201610406700645e-05, + "loss": 0.0131, + "step": 6010 + }, + { + "epoch": 10.290598290598291, + "grad_norm": 0.07714977115392685, + "learning_rate": 1.6160401997021406e-05, + "loss": 0.0148, + "step": 6020 + }, + { + "epoch": 10.307692307692308, + "grad_norm": 0.07542186975479126, + "learning_rate": 1.6119184774860322e-05, + "loss": 0.0151, + "step": 6030 + }, + { + "epoch": 10.324786324786325, + "grad_norm": 0.06268887966871262, + "learning_rate": 1.6077959053234796e-05, + "loss": 0.0143, + "step": 6040 + }, + { + "epoch": 10.341880341880342, + "grad_norm": 0.06660132855176926, + "learning_rate": 1.6036725145226774e-05, + "loss": 0.0118, + "step": 6050 + }, + { + "epoch": 10.35897435897436, + "grad_norm": 0.09458441287279129, + "learning_rate": 1.5995483363980394e-05, + "loss": 0.0136, + "step": 6060 + }, + { + "epoch": 10.376068376068377, + "grad_norm": 0.09146513044834137, + "learning_rate": 1.595423402269956e-05, + "loss": 0.0138, + "step": 6070 + }, + { + "epoch": 10.393162393162394, + "grad_norm": 0.0746873989701271, + "learning_rate": 1.5912977434645608e-05, + "loss": 0.0161, + "step": 6080 + }, + { + "epoch": 10.41025641025641, + "grad_norm": 0.07004496455192566, + "learning_rate": 1.587171391313489e-05, + "loss": 0.0136, + "step": 6090 + }, + { + "epoch": 10.427350427350428, + "grad_norm": 0.059027332812547684, + "learning_rate": 1.5830443771536434e-05, + "loss": 0.0142, + "step": 6100 + }, + { + "epoch": 10.444444444444445, + "grad_norm": 0.08229991793632507, + "learning_rate": 1.578916732326953e-05, + "loss": 0.0153, + "step": 6110 + }, + { + "epoch": 10.461538461538462, + "grad_norm": 0.1089404970407486, + "learning_rate": 1.5747884881801364e-05, + "loss": 0.0139, + "step": 6120 + }, + { + "epoch": 10.478632478632479, + "grad_norm": 0.07712269574403763, + "learning_rate": 1.570659676064464e-05, + "loss": 0.0147, + "step": 6130 + }, + { + "epoch": 10.495726495726496, + "grad_norm": 0.06837616115808487, + "learning_rate": 1.5665303273355197e-05, + "loss": 0.0141, + "step": 6140 + }, + { + "epoch": 10.512820512820513, + "grad_norm": 0.06299047917127609, + "learning_rate": 1.562400473352962e-05, + "loss": 0.0132, + "step": 6150 + }, + { + "epoch": 10.52991452991453, + "grad_norm": 0.09000428766012192, + "learning_rate": 1.5582701454802867e-05, + "loss": 0.0118, + "step": 6160 + }, + { + "epoch": 10.547008547008547, + "grad_norm": 0.08584591001272202, + "learning_rate": 1.5541393750845886e-05, + "loss": 0.0119, + "step": 6170 + }, + { + "epoch": 10.564102564102564, + "grad_norm": 0.06676366180181503, + "learning_rate": 1.5500081935363233e-05, + "loss": 0.0128, + "step": 6180 + }, + { + "epoch": 10.581196581196581, + "grad_norm": 0.0854257270693779, + "learning_rate": 1.5458766322090684e-05, + "loss": 0.0144, + "step": 6190 + }, + { + "epoch": 10.598290598290598, + "grad_norm": 0.07970619201660156, + "learning_rate": 1.5417447224792862e-05, + "loss": 0.0131, + "step": 6200 + }, + { + "epoch": 10.615384615384615, + "grad_norm": 0.10973725467920303, + "learning_rate": 1.5376124957260847e-05, + "loss": 0.015, + "step": 6210 + }, + { + "epoch": 10.632478632478632, + "grad_norm": 0.08757691085338593, + "learning_rate": 1.5334799833309792e-05, + "loss": 0.0143, + "step": 6220 + }, + { + "epoch": 10.649572649572649, + "grad_norm": 0.0727427676320076, + "learning_rate": 1.529347216677654e-05, + "loss": 0.0131, + "step": 6230 + }, + { + "epoch": 10.666666666666666, + "grad_norm": 0.07811034470796585, + "learning_rate": 1.5252142271517259e-05, + "loss": 0.0141, + "step": 6240 + }, + { + "epoch": 10.683760683760683, + "grad_norm": 0.0741906389594078, + "learning_rate": 1.5210810461405019e-05, + "loss": 0.0122, + "step": 6250 + }, + { + "epoch": 10.7008547008547, + "grad_norm": 0.10513097047805786, + "learning_rate": 1.5169477050327457e-05, + "loss": 0.0128, + "step": 6260 + }, + { + "epoch": 10.717948717948717, + "grad_norm": 0.1017889752984047, + "learning_rate": 1.5128142352184346e-05, + "loss": 0.0126, + "step": 6270 + }, + { + "epoch": 10.735042735042736, + "grad_norm": 0.07930806279182434, + "learning_rate": 1.5086806680885245e-05, + "loss": 0.0144, + "step": 6280 + }, + { + "epoch": 10.752136752136753, + "grad_norm": 0.07535891234874725, + "learning_rate": 1.5045470350347106e-05, + "loss": 0.0135, + "step": 6290 + }, + { + "epoch": 10.76923076923077, + "grad_norm": 0.09707707911729813, + "learning_rate": 1.5004133674491876e-05, + "loss": 0.0123, + "step": 6300 + }, + { + "epoch": 10.786324786324787, + "grad_norm": 0.07067103683948517, + "learning_rate": 1.4962796967244137e-05, + "loss": 0.0132, + "step": 6310 + }, + { + "epoch": 10.803418803418804, + "grad_norm": 0.0870518609881401, + "learning_rate": 1.4921460542528706e-05, + "loss": 0.0134, + "step": 6320 + }, + { + "epoch": 10.820512820512821, + "grad_norm": 0.08242181688547134, + "learning_rate": 1.4880124714268248e-05, + "loss": 0.0133, + "step": 6330 + }, + { + "epoch": 10.837606837606838, + "grad_norm": 0.09291113913059235, + "learning_rate": 1.4838789796380901e-05, + "loss": 0.0148, + "step": 6340 + }, + { + "epoch": 10.854700854700855, + "grad_norm": 0.0815432071685791, + "learning_rate": 1.4797456102777899e-05, + "loss": 0.0145, + "step": 6350 + }, + { + "epoch": 10.871794871794872, + "grad_norm": 0.0877620130777359, + "learning_rate": 1.4756123947361164e-05, + "loss": 0.0129, + "step": 6360 + }, + { + "epoch": 10.88888888888889, + "grad_norm": 0.09172259271144867, + "learning_rate": 1.4714793644020942e-05, + "loss": 0.0141, + "step": 6370 + }, + { + "epoch": 10.905982905982906, + "grad_norm": 0.07486597448587418, + "learning_rate": 1.4673465506633417e-05, + "loss": 0.0132, + "step": 6380 + }, + { + "epoch": 10.923076923076923, + "grad_norm": 0.0865323394536972, + "learning_rate": 1.4632139849058328e-05, + "loss": 0.0135, + "step": 6390 + }, + { + "epoch": 10.94017094017094, + "grad_norm": 0.09112763404846191, + "learning_rate": 1.4590816985136563e-05, + "loss": 0.0145, + "step": 6400 + }, + { + "epoch": 10.957264957264957, + "grad_norm": 0.07164663076400757, + "learning_rate": 1.4549497228687812e-05, + "loss": 0.0145, + "step": 6410 + }, + { + "epoch": 10.974358974358974, + "grad_norm": 0.07041939347982407, + "learning_rate": 1.4508180893508157e-05, + "loss": 0.0142, + "step": 6420 + }, + { + "epoch": 10.991452991452991, + "grad_norm": 0.0761607214808464, + "learning_rate": 1.4466868293367708e-05, + "loss": 0.012, + "step": 6430 + }, + { + "epoch": 11.008547008547009, + "grad_norm": 0.09176713228225708, + "learning_rate": 1.4425559742008196e-05, + "loss": 0.014, + "step": 6440 + }, + { + "epoch": 11.025641025641026, + "grad_norm": 0.08743807673454285, + "learning_rate": 1.438425555314061e-05, + "loss": 0.0139, + "step": 6450 + }, + { + "epoch": 11.042735042735043, + "grad_norm": 0.08176425844430923, + "learning_rate": 1.4342956040442816e-05, + "loss": 0.0133, + "step": 6460 + }, + { + "epoch": 11.05982905982906, + "grad_norm": 0.07121485471725464, + "learning_rate": 1.4301661517557155e-05, + "loss": 0.0128, + "step": 6470 + }, + { + "epoch": 11.076923076923077, + "grad_norm": 0.07166894525289536, + "learning_rate": 1.426037229808808e-05, + "loss": 0.014, + "step": 6480 + }, + { + "epoch": 11.094017094017094, + "grad_norm": 0.08936766535043716, + "learning_rate": 1.4219088695599773e-05, + "loss": 0.0129, + "step": 6490 + }, + { + "epoch": 11.11111111111111, + "grad_norm": 0.0939464345574379, + "learning_rate": 1.4177811023613749e-05, + "loss": 0.0134, + "step": 6500 + }, + { + "epoch": 11.128205128205128, + "grad_norm": 0.07643920928239822, + "learning_rate": 1.4136539595606493e-05, + "loss": 0.015, + "step": 6510 + }, + { + "epoch": 11.145299145299145, + "grad_norm": 0.08625482022762299, + "learning_rate": 1.4095274725007064e-05, + "loss": 0.014, + "step": 6520 + }, + { + "epoch": 11.162393162393162, + "grad_norm": 0.06937939673662186, + "learning_rate": 1.4054016725194727e-05, + "loss": 0.0111, + "step": 6530 + }, + { + "epoch": 11.179487179487179, + "grad_norm": 0.058906346559524536, + "learning_rate": 1.4012765909496567e-05, + "loss": 0.012, + "step": 6540 + }, + { + "epoch": 11.196581196581196, + "grad_norm": 0.08350737392902374, + "learning_rate": 1.3971522591185111e-05, + "loss": 0.0125, + "step": 6550 + }, + { + "epoch": 11.213675213675213, + "grad_norm": 0.08366569876670837, + "learning_rate": 1.3930287083475943e-05, + "loss": 0.0126, + "step": 6560 + }, + { + "epoch": 11.23076923076923, + "grad_norm": 0.07450778037309647, + "learning_rate": 1.3889059699525338e-05, + "loss": 0.0138, + "step": 6570 + }, + { + "epoch": 11.247863247863247, + "grad_norm": 0.08065537363290787, + "learning_rate": 1.3847840752427874e-05, + "loss": 0.0118, + "step": 6580 + }, + { + "epoch": 11.264957264957266, + "grad_norm": 0.08331799507141113, + "learning_rate": 1.380663055521405e-05, + "loss": 0.0131, + "step": 6590 + }, + { + "epoch": 11.282051282051283, + "grad_norm": 0.061685334891080856, + "learning_rate": 1.3765429420847928e-05, + "loss": 0.0118, + "step": 6600 + }, + { + "epoch": 11.2991452991453, + "grad_norm": 0.0562916062772274, + "learning_rate": 1.3724237662224738e-05, + "loss": 0.0145, + "step": 6610 + }, + { + "epoch": 11.316239316239317, + "grad_norm": 0.06560176610946655, + "learning_rate": 1.3683055592168497e-05, + "loss": 0.0132, + "step": 6620 + }, + { + "epoch": 11.333333333333334, + "grad_norm": 0.07403317838907242, + "learning_rate": 1.3641883523429665e-05, + "loss": 0.0138, + "step": 6630 + }, + { + "epoch": 11.350427350427351, + "grad_norm": 0.07964484393596649, + "learning_rate": 1.3600721768682729e-05, + "loss": 0.0161, + "step": 6640 + }, + { + "epoch": 11.367521367521368, + "grad_norm": 0.0782715305685997, + "learning_rate": 1.3559570640523858e-05, + "loss": 0.0155, + "step": 6650 + }, + { + "epoch": 11.384615384615385, + "grad_norm": 0.06219352409243584, + "learning_rate": 1.351843045146852e-05, + "loss": 0.0112, + "step": 6660 + }, + { + "epoch": 11.401709401709402, + "grad_norm": 0.07144197076559067, + "learning_rate": 1.3477301513949103e-05, + "loss": 0.0142, + "step": 6670 + }, + { + "epoch": 11.418803418803419, + "grad_norm": 0.07987230271100998, + "learning_rate": 1.3436184140312546e-05, + "loss": 0.012, + "step": 6680 + }, + { + "epoch": 11.435897435897436, + "grad_norm": 0.07232145965099335, + "learning_rate": 1.3395078642817976e-05, + "loss": 0.012, + "step": 6690 + }, + { + "epoch": 11.452991452991453, + "grad_norm": 0.09328833967447281, + "learning_rate": 1.3353985333634321e-05, + "loss": 0.0123, + "step": 6700 + }, + { + "epoch": 11.47008547008547, + "grad_norm": 0.05845474824309349, + "learning_rate": 1.3312904524837948e-05, + "loss": 0.0128, + "step": 6710 + }, + { + "epoch": 11.487179487179487, + "grad_norm": 0.08848857879638672, + "learning_rate": 1.3271836528410299e-05, + "loss": 0.0147, + "step": 6720 + }, + { + "epoch": 11.504273504273504, + "grad_norm": 0.08257392048835754, + "learning_rate": 1.3230781656235505e-05, + "loss": 0.0128, + "step": 6730 + }, + { + "epoch": 11.521367521367521, + "grad_norm": 0.07344449311494827, + "learning_rate": 1.3189740220098026e-05, + "loss": 0.0123, + "step": 6740 + }, + { + "epoch": 11.538461538461538, + "grad_norm": 0.0760531798005104, + "learning_rate": 1.31487125316803e-05, + "loss": 0.0157, + "step": 6750 + }, + { + "epoch": 11.555555555555555, + "grad_norm": 0.064517080783844, + "learning_rate": 1.3107698902560344e-05, + "loss": 0.0133, + "step": 6760 + }, + { + "epoch": 11.572649572649572, + "grad_norm": 0.05018634349107742, + "learning_rate": 1.3066699644209404e-05, + "loss": 0.0122, + "step": 6770 + }, + { + "epoch": 11.58974358974359, + "grad_norm": 0.07061044126749039, + "learning_rate": 1.3025715067989606e-05, + "loss": 0.0148, + "step": 6780 + }, + { + "epoch": 11.606837606837606, + "grad_norm": 0.1009698361158371, + "learning_rate": 1.2984745485151559e-05, + "loss": 0.013, + "step": 6790 + }, + { + "epoch": 11.623931623931623, + "grad_norm": 0.06724962592124939, + "learning_rate": 1.2943791206832008e-05, + "loss": 0.0133, + "step": 6800 + }, + { + "epoch": 11.64102564102564, + "grad_norm": 0.0638575479388237, + "learning_rate": 1.290285254405148e-05, + "loss": 0.0126, + "step": 6810 + }, + { + "epoch": 11.658119658119658, + "grad_norm": 0.07809091359376907, + "learning_rate": 1.2861929807711901e-05, + "loss": 0.0146, + "step": 6820 + }, + { + "epoch": 11.675213675213675, + "grad_norm": 0.08215625584125519, + "learning_rate": 1.2821023308594256e-05, + "loss": 0.0135, + "step": 6830 + }, + { + "epoch": 11.692307692307692, + "grad_norm": 0.07283703237771988, + "learning_rate": 1.2780133357356218e-05, + "loss": 0.0128, + "step": 6840 + }, + { + "epoch": 11.709401709401709, + "grad_norm": 0.07699073851108551, + "learning_rate": 1.273926026452978e-05, + "loss": 0.0128, + "step": 6850 + }, + { + "epoch": 11.726495726495726, + "grad_norm": 0.06167527660727501, + "learning_rate": 1.2698404340518922e-05, + "loss": 0.0125, + "step": 6860 + }, + { + "epoch": 11.743589743589745, + "grad_norm": 0.06687825918197632, + "learning_rate": 1.2657565895597222e-05, + "loss": 0.0116, + "step": 6870 + }, + { + "epoch": 11.760683760683762, + "grad_norm": 0.07788685709238052, + "learning_rate": 1.2616745239905527e-05, + "loss": 0.0143, + "step": 6880 + }, + { + "epoch": 11.777777777777779, + "grad_norm": 0.11028165370225906, + "learning_rate": 1.2575942683449585e-05, + "loss": 0.013, + "step": 6890 + }, + { + "epoch": 11.794871794871796, + "grad_norm": 0.08098877966403961, + "learning_rate": 1.2535158536097691e-05, + "loss": 0.0121, + "step": 6900 + }, + { + "epoch": 11.811965811965813, + "grad_norm": 0.061440546065568924, + "learning_rate": 1.249439310757833e-05, + "loss": 0.0132, + "step": 6910 + }, + { + "epoch": 11.82905982905983, + "grad_norm": 0.06367921084165573, + "learning_rate": 1.2453646707477843e-05, + "loss": 0.0134, + "step": 6920 + }, + { + "epoch": 11.846153846153847, + "grad_norm": 0.07408096641302109, + "learning_rate": 1.2412919645238049e-05, + "loss": 0.0116, + "step": 6930 + }, + { + "epoch": 11.863247863247864, + "grad_norm": 0.05494624003767967, + "learning_rate": 1.2372212230153911e-05, + "loss": 0.0121, + "step": 6940 + }, + { + "epoch": 11.88034188034188, + "grad_norm": 0.08546881377696991, + "learning_rate": 1.2331524771371197e-05, + "loss": 0.0128, + "step": 6950 + }, + { + "epoch": 11.897435897435898, + "grad_norm": 0.06938689202070236, + "learning_rate": 1.2290857577884109e-05, + "loss": 0.0131, + "step": 6960 + }, + { + "epoch": 11.914529914529915, + "grad_norm": 0.09075794368982315, + "learning_rate": 1.2250210958532943e-05, + "loss": 0.012, + "step": 6970 + }, + { + "epoch": 11.931623931623932, + "grad_norm": 0.06405825167894363, + "learning_rate": 1.2209585222001762e-05, + "loss": 0.0137, + "step": 6980 + }, + { + "epoch": 11.948717948717949, + "grad_norm": 0.097632996737957, + "learning_rate": 1.216898067681603e-05, + "loss": 0.0123, + "step": 6990 + }, + { + "epoch": 11.965811965811966, + "grad_norm": 0.08975306153297424, + "learning_rate": 1.2128397631340274e-05, + "loss": 0.0121, + "step": 7000 + }, + { + "epoch": 11.982905982905983, + "grad_norm": 0.08020646870136261, + "learning_rate": 1.2087836393775756e-05, + "loss": 0.013, + "step": 7010 + }, + { + "epoch": 12.0, + "grad_norm": 0.08891727030277252, + "learning_rate": 1.2047297272158112e-05, + "loss": 0.0127, + "step": 7020 + }, + { + "epoch": 12.017094017094017, + "grad_norm": 0.0835188701748848, + "learning_rate": 1.2006780574355017e-05, + "loss": 0.0113, + "step": 7030 + }, + { + "epoch": 12.034188034188034, + "grad_norm": 0.0732852891087532, + "learning_rate": 1.1966286608063866e-05, + "loss": 0.0124, + "step": 7040 + }, + { + "epoch": 12.051282051282051, + "grad_norm": 0.07198654860258102, + "learning_rate": 1.1925815680809419e-05, + "loss": 0.0108, + "step": 7050 + }, + { + "epoch": 12.068376068376068, + "grad_norm": 0.08659420907497406, + "learning_rate": 1.1885368099941459e-05, + "loss": 0.0127, + "step": 7060 + }, + { + "epoch": 12.085470085470085, + "grad_norm": 0.06744220107793808, + "learning_rate": 1.1844944172632485e-05, + "loss": 0.0124, + "step": 7070 + }, + { + "epoch": 12.102564102564102, + "grad_norm": 0.07504130899906158, + "learning_rate": 1.1804544205875353e-05, + "loss": 0.0144, + "step": 7080 + }, + { + "epoch": 12.11965811965812, + "grad_norm": 0.08647613972425461, + "learning_rate": 1.1764168506480953e-05, + "loss": 0.0115, + "step": 7090 + }, + { + "epoch": 12.136752136752136, + "grad_norm": 0.06630230695009232, + "learning_rate": 1.1723817381075889e-05, + "loss": 0.0122, + "step": 7100 + }, + { + "epoch": 12.153846153846153, + "grad_norm": 0.08289078623056412, + "learning_rate": 1.1683491136100135e-05, + "loss": 0.0125, + "step": 7110 + }, + { + "epoch": 12.17094017094017, + "grad_norm": 0.07730842381715775, + "learning_rate": 1.1643190077804713e-05, + "loss": 0.0132, + "step": 7120 + }, + { + "epoch": 12.188034188034187, + "grad_norm": 0.060852617025375366, + "learning_rate": 1.1602914512249379e-05, + "loss": 0.0132, + "step": 7130 + }, + { + "epoch": 12.205128205128204, + "grad_norm": 0.08456648141145706, + "learning_rate": 1.156266474530028e-05, + "loss": 0.0105, + "step": 7140 + }, + { + "epoch": 12.222222222222221, + "grad_norm": 0.07040031999349594, + "learning_rate": 1.152244108262764e-05, + "loss": 0.0103, + "step": 7150 + }, + { + "epoch": 12.239316239316238, + "grad_norm": 0.05960635840892792, + "learning_rate": 1.1482243829703444e-05, + "loss": 0.0109, + "step": 7160 + }, + { + "epoch": 12.256410256410255, + "grad_norm": 0.06489500403404236, + "learning_rate": 1.1442073291799106e-05, + "loss": 0.0134, + "step": 7170 + }, + { + "epoch": 12.273504273504274, + "grad_norm": 0.09831108152866364, + "learning_rate": 1.1401929773983155e-05, + "loss": 0.0138, + "step": 7180 + }, + { + "epoch": 12.290598290598291, + "grad_norm": 0.06196901202201843, + "learning_rate": 1.1361813581118936e-05, + "loss": 0.0137, + "step": 7190 + }, + { + "epoch": 12.307692307692308, + "grad_norm": 0.067090705037117, + "learning_rate": 1.1321725017862268e-05, + "loss": 0.0118, + "step": 7200 + }, + { + "epoch": 12.324786324786325, + "grad_norm": 0.06177959218621254, + "learning_rate": 1.1281664388659134e-05, + "loss": 0.0116, + "step": 7210 + }, + { + "epoch": 12.341880341880342, + "grad_norm": 0.08829686790704727, + "learning_rate": 1.1241631997743384e-05, + "loss": 0.0099, + "step": 7220 + }, + { + "epoch": 12.35897435897436, + "grad_norm": 0.07227877527475357, + "learning_rate": 1.1201628149134422e-05, + "loss": 0.0131, + "step": 7230 + }, + { + "epoch": 12.376068376068377, + "grad_norm": 0.06643012911081314, + "learning_rate": 1.116165314663489e-05, + "loss": 0.0128, + "step": 7240 + }, + { + "epoch": 12.393162393162394, + "grad_norm": 0.08133196830749512, + "learning_rate": 1.1121707293828353e-05, + "loss": 0.0114, + "step": 7250 + }, + { + "epoch": 12.41025641025641, + "grad_norm": 0.0630437433719635, + "learning_rate": 1.1081790894077022e-05, + "loss": 0.0117, + "step": 7260 + }, + { + "epoch": 12.427350427350428, + "grad_norm": 0.08086523413658142, + "learning_rate": 1.104190425051941e-05, + "loss": 0.0139, + "step": 7270 + }, + { + "epoch": 12.444444444444445, + "grad_norm": 0.0775720551609993, + "learning_rate": 1.1002047666068066e-05, + "loss": 0.0132, + "step": 7280 + }, + { + "epoch": 12.461538461538462, + "grad_norm": 0.08362945914268494, + "learning_rate": 1.0962221443407257e-05, + "loss": 0.0129, + "step": 7290 + }, + { + "epoch": 12.478632478632479, + "grad_norm": 0.06774546205997467, + "learning_rate": 1.0922425884990673e-05, + "loss": 0.0121, + "step": 7300 + }, + { + "epoch": 12.495726495726496, + "grad_norm": 0.06199425831437111, + "learning_rate": 1.0882661293039124e-05, + "loss": 0.0111, + "step": 7310 + }, + { + "epoch": 12.512820512820513, + "grad_norm": 0.1027790978550911, + "learning_rate": 1.0842927969538258e-05, + "loss": 0.0136, + "step": 7320 + }, + { + "epoch": 12.52991452991453, + "grad_norm": 0.09067315608263016, + "learning_rate": 1.0803226216236256e-05, + "loss": 0.0126, + "step": 7330 + }, + { + "epoch": 12.547008547008547, + "grad_norm": 0.08140069246292114, + "learning_rate": 1.076355633464154e-05, + "loss": 0.0121, + "step": 7340 + }, + { + "epoch": 12.564102564102564, + "grad_norm": 0.08362126350402832, + "learning_rate": 1.0723918626020498e-05, + "loss": 0.0119, + "step": 7350 + }, + { + "epoch": 12.581196581196581, + "grad_norm": 0.07784171402454376, + "learning_rate": 1.0684313391395176e-05, + "loss": 0.0134, + "step": 7360 + }, + { + "epoch": 12.598290598290598, + "grad_norm": 0.05226428434252739, + "learning_rate": 1.0644740931541005e-05, + "loss": 0.0139, + "step": 7370 + }, + { + "epoch": 12.615384615384615, + "grad_norm": 0.07026848196983337, + "learning_rate": 1.0605201546984517e-05, + "loss": 0.0132, + "step": 7380 + }, + { + "epoch": 12.632478632478632, + "grad_norm": 0.07804906368255615, + "learning_rate": 1.0565695538001056e-05, + "loss": 0.0158, + "step": 7390 + }, + { + "epoch": 12.649572649572649, + "grad_norm": 0.07104866206645966, + "learning_rate": 1.0526223204612495e-05, + "loss": 0.0122, + "step": 7400 + }, + { + "epoch": 12.666666666666666, + "grad_norm": 0.09126023948192596, + "learning_rate": 1.0486784846584974e-05, + "loss": 0.0114, + "step": 7410 + }, + { + "epoch": 12.683760683760683, + "grad_norm": 0.09580039232969284, + "learning_rate": 1.0447380763426607e-05, + "loss": 0.0122, + "step": 7420 + }, + { + "epoch": 12.7008547008547, + "grad_norm": 0.06293272227048874, + "learning_rate": 1.040801125438521e-05, + "loss": 0.0116, + "step": 7430 + }, + { + "epoch": 12.717948717948717, + "grad_norm": 0.08728829026222229, + "learning_rate": 1.0368676618446042e-05, + "loss": 0.0112, + "step": 7440 + }, + { + "epoch": 12.735042735042736, + "grad_norm": 0.07772357761859894, + "learning_rate": 1.0329377154329512e-05, + "loss": 0.0126, + "step": 7450 + }, + { + "epoch": 12.752136752136753, + "grad_norm": 0.15152858197689056, + "learning_rate": 1.0290113160488925e-05, + "loss": 0.0107, + "step": 7460 + }, + { + "epoch": 12.76923076923077, + "grad_norm": 0.07972808182239532, + "learning_rate": 1.0250884935108223e-05, + "loss": 0.0128, + "step": 7470 + }, + { + "epoch": 12.786324786324787, + "grad_norm": 0.08153508603572845, + "learning_rate": 1.0211692776099698e-05, + "loss": 0.0121, + "step": 7480 + }, + { + "epoch": 12.803418803418804, + "grad_norm": 0.07217764109373093, + "learning_rate": 1.0172536981101747e-05, + "loss": 0.0111, + "step": 7490 + }, + { + "epoch": 12.820512820512821, + "grad_norm": 0.06588239222764969, + "learning_rate": 1.0133417847476606e-05, + "loss": 0.012, + "step": 7500 + }, + { + "epoch": 12.837606837606838, + "grad_norm": 0.09637109935283661, + "learning_rate": 1.0094335672308095e-05, + "loss": 0.0113, + "step": 7510 + }, + { + "epoch": 12.854700854700855, + "grad_norm": 0.05825330689549446, + "learning_rate": 1.0055290752399353e-05, + "loss": 0.0126, + "step": 7520 + }, + { + "epoch": 12.871794871794872, + "grad_norm": 0.07299874722957611, + "learning_rate": 1.0016283384270594e-05, + "loss": 0.0109, + "step": 7530 + }, + { + "epoch": 12.88888888888889, + "grad_norm": 0.0607651025056839, + "learning_rate": 9.977313864156854e-06, + "loss": 0.0125, + "step": 7540 + }, + { + "epoch": 12.905982905982906, + "grad_norm": 0.08556436747312546, + "learning_rate": 9.938382488005729e-06, + "loss": 0.0106, + "step": 7550 + }, + { + "epoch": 12.923076923076923, + "grad_norm": 0.054035484790802, + "learning_rate": 9.89948955147515e-06, + "loss": 0.0136, + "step": 7560 + }, + { + "epoch": 12.94017094017094, + "grad_norm": 0.06668644398450851, + "learning_rate": 9.860635349931116e-06, + "loss": 0.0112, + "step": 7570 + }, + { + "epoch": 12.957264957264957, + "grad_norm": 0.07119438052177429, + "learning_rate": 9.821820178445455e-06, + "loss": 0.0113, + "step": 7580 + }, + { + "epoch": 12.974358974358974, + "grad_norm": 0.0803217887878418, + "learning_rate": 9.783044331793605e-06, + "loss": 0.0107, + "step": 7590 + }, + { + "epoch": 12.991452991452991, + "grad_norm": 0.062193505465984344, + "learning_rate": 9.744308104452338e-06, + "loss": 0.0119, + "step": 7600 + }, + { + "epoch": 13.008547008547009, + "grad_norm": 0.08492100983858109, + "learning_rate": 9.705611790597562e-06, + "loss": 0.0114, + "step": 7610 + }, + { + "epoch": 13.025641025641026, + "grad_norm": 0.06370862573385239, + "learning_rate": 9.666955684102043e-06, + "loss": 0.0116, + "step": 7620 + }, + { + "epoch": 13.042735042735043, + "grad_norm": 0.05765674635767937, + "learning_rate": 9.62834007853323e-06, + "loss": 0.0122, + "step": 7630 + }, + { + "epoch": 13.05982905982906, + "grad_norm": 0.05570053681731224, + "learning_rate": 9.589765267150975e-06, + "loss": 0.0127, + "step": 7640 + }, + { + "epoch": 13.076923076923077, + "grad_norm": 0.04684542492032051, + "learning_rate": 9.551231542905326e-06, + "loss": 0.0126, + "step": 7650 + }, + { + "epoch": 13.094017094017094, + "grad_norm": 0.10144971311092377, + "learning_rate": 9.512739198434312e-06, + "loss": 0.0131, + "step": 7660 + }, + { + "epoch": 13.11111111111111, + "grad_norm": 0.06794734299182892, + "learning_rate": 9.474288526061698e-06, + "loss": 0.0148, + "step": 7670 + }, + { + "epoch": 13.128205128205128, + "grad_norm": 0.06705810874700546, + "learning_rate": 9.435879817794784e-06, + "loss": 0.0115, + "step": 7680 + }, + { + "epoch": 13.145299145299145, + "grad_norm": 0.06749339401721954, + "learning_rate": 9.39751336532218e-06, + "loss": 0.0097, + "step": 7690 + }, + { + "epoch": 13.162393162393162, + "grad_norm": 0.07151390612125397, + "learning_rate": 9.359189460011591e-06, + "loss": 0.014, + "step": 7700 + }, + { + "epoch": 13.179487179487179, + "grad_norm": 0.05107467994093895, + "learning_rate": 9.3209083929076e-06, + "loss": 0.0137, + "step": 7710 + }, + { + "epoch": 13.196581196581196, + "grad_norm": 0.0684099793434143, + "learning_rate": 9.282670454729471e-06, + "loss": 0.01, + "step": 7720 + }, + { + "epoch": 13.213675213675213, + "grad_norm": 0.1151367649435997, + "learning_rate": 9.244475935868927e-06, + "loss": 0.0137, + "step": 7730 + }, + { + "epoch": 13.23076923076923, + "grad_norm": 0.09851977974176407, + "learning_rate": 9.206325126387945e-06, + "loss": 0.0117, + "step": 7740 + }, + { + "epoch": 13.247863247863247, + "grad_norm": 0.05303683504462242, + "learning_rate": 9.168218316016572e-06, + "loss": 0.0116, + "step": 7750 + }, + { + "epoch": 13.264957264957266, + "grad_norm": 0.08101697266101837, + "learning_rate": 9.130155794150698e-06, + "loss": 0.0116, + "step": 7760 + }, + { + "epoch": 13.282051282051283, + "grad_norm": 0.07205120474100113, + "learning_rate": 9.092137849849876e-06, + "loss": 0.014, + "step": 7770 + }, + { + "epoch": 13.2991452991453, + "grad_norm": 0.10680156201124191, + "learning_rate": 9.054164771835128e-06, + "loss": 0.0122, + "step": 7780 + }, + { + "epoch": 13.316239316239317, + "grad_norm": 0.07473104447126389, + "learning_rate": 9.016236848486736e-06, + "loss": 0.0106, + "step": 7790 + }, + { + "epoch": 13.333333333333334, + "grad_norm": 0.09170293807983398, + "learning_rate": 8.978354367842068e-06, + "loss": 0.0113, + "step": 7800 + }, + { + "epoch": 13.350427350427351, + "grad_norm": 0.10518091171979904, + "learning_rate": 8.940517617593384e-06, + "loss": 0.0105, + "step": 7810 + }, + { + "epoch": 13.367521367521368, + "grad_norm": 0.08690715581178665, + "learning_rate": 8.902726885085655e-06, + "loss": 0.0084, + "step": 7820 + }, + { + "epoch": 13.384615384615385, + "grad_norm": 0.08386173844337463, + "learning_rate": 8.864982457314365e-06, + "loss": 0.0134, + "step": 7830 + }, + { + "epoch": 13.401709401709402, + "grad_norm": 0.06394433975219727, + "learning_rate": 8.827284620923369e-06, + "loss": 0.0111, + "step": 7840 + }, + { + "epoch": 13.418803418803419, + "grad_norm": 0.0801367238163948, + "learning_rate": 8.789633662202667e-06, + "loss": 0.0115, + "step": 7850 + }, + { + "epoch": 13.435897435897436, + "grad_norm": 0.07522069662809372, + "learning_rate": 8.752029867086264e-06, + "loss": 0.0126, + "step": 7860 + }, + { + "epoch": 13.452991452991453, + "grad_norm": 0.06898823380470276, + "learning_rate": 8.714473521149994e-06, + "loss": 0.0113, + "step": 7870 + }, + { + "epoch": 13.47008547008547, + "grad_norm": 0.07964406907558441, + "learning_rate": 8.676964909609334e-06, + "loss": 0.0119, + "step": 7880 + }, + { + "epoch": 13.487179487179487, + "grad_norm": 0.08102905005216599, + "learning_rate": 8.639504317317258e-06, + "loss": 0.0115, + "step": 7890 + }, + { + "epoch": 13.504273504273504, + "grad_norm": 0.0762372612953186, + "learning_rate": 8.602092028762073e-06, + "loss": 0.0128, + "step": 7900 + }, + { + "epoch": 13.521367521367521, + "grad_norm": 0.07291470468044281, + "learning_rate": 8.56472832806523e-06, + "loss": 0.0115, + "step": 7910 + }, + { + "epoch": 13.538461538461538, + "grad_norm": 0.08234073221683502, + "learning_rate": 8.527413498979204e-06, + "loss": 0.0137, + "step": 7920 + }, + { + "epoch": 13.555555555555555, + "grad_norm": 0.05066027492284775, + "learning_rate": 8.490147824885328e-06, + "loss": 0.0117, + "step": 7930 + }, + { + "epoch": 13.572649572649572, + "grad_norm": 0.11162367463111877, + "learning_rate": 8.452931588791614e-06, + "loss": 0.0114, + "step": 7940 + }, + { + "epoch": 13.58974358974359, + "grad_norm": 0.06701509654521942, + "learning_rate": 8.415765073330642e-06, + "loss": 0.0123, + "step": 7950 + }, + { + "epoch": 13.606837606837606, + "grad_norm": 0.06369341164827347, + "learning_rate": 8.378648560757396e-06, + "loss": 0.0123, + "step": 7960 + }, + { + "epoch": 13.623931623931623, + "grad_norm": 0.061316028237342834, + "learning_rate": 8.34158233294711e-06, + "loss": 0.0148, + "step": 7970 + }, + { + "epoch": 13.64102564102564, + "grad_norm": 0.08117634057998657, + "learning_rate": 8.304566671393151e-06, + "loss": 0.012, + "step": 7980 + }, + { + "epoch": 13.658119658119658, + "grad_norm": 0.08008228242397308, + "learning_rate": 8.267601857204869e-06, + "loss": 0.0133, + "step": 7990 + }, + { + "epoch": 13.675213675213675, + "grad_norm": 0.05691583827137947, + "learning_rate": 8.230688171105446e-06, + "loss": 0.011, + "step": 8000 + } + ], + "logging_steps": 10, + "max_steps": 12000, + "num_input_tokens_seen": 0, + "num_train_epochs": 21, + "save_steps": 12001, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +} diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/CheesyBread/episode_results.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/CheesyBread/episode_results.json new file mode 100644 index 0000000000000000000000000000000000000000..208545a21541e5b8f665c0ccbdc75f108db73a15 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/CheesyBread/episode_results.json @@ -0,0 +1,602 @@ +[ + { + "completion_order": 0, + "task_name": "CheesyBread", + "episode_idx": 0, + "seed": 42, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 1, + "task_name": "CheesyBread", + "episode_idx": 1, + "seed": 43, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 16.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 2, + "task_name": "CheesyBread", + "episode_idx": 2, + "seed": 44, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 25.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 3, + "task_name": "CheesyBread", + "episode_idx": 3, + "seed": 45, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 4, + "task_name": "CheesyBread", + "episode_idx": 4, + "seed": 46, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 5, + "task_name": "CheesyBread", + "episode_idx": 5, + "seed": 47, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 6, + "task_name": "CheesyBread", + "episode_idx": 6, + "seed": 48, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 7, + "task_name": "CheesyBread", + "episode_idx": 7, + "seed": 49, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 8, + "task_name": "CheesyBread", + "episode_idx": 8, + "seed": 50, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 27.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 9, + "task_name": "CheesyBread", + "episode_idx": 9, + "seed": 51, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 33.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 10, + "task_name": "CheesyBread", + "episode_idx": 10, + "seed": 52, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 11, + "task_name": "CheesyBread", + "episode_idx": 11, + "seed": 53, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 12, + "task_name": "CheesyBread", + "episode_idx": 12, + "seed": 54, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 29.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 13, + "task_name": "CheesyBread", + "episode_idx": 13, + "seed": 55, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 14, + "task_name": "CheesyBread", + "episode_idx": 14, + "seed": 56, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 15, + "task_name": "CheesyBread", + "episode_idx": 15, + "seed": 57, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 25.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 16, + "task_name": "CheesyBread", + "episode_idx": 16, + "seed": 58, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 17, + "task_name": "CheesyBread", + "episode_idx": 17, + "seed": 59, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 18, + "task_name": "CheesyBread", + "episode_idx": 18, + "seed": 60, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 33.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 19, + "task_name": "CheesyBread", + "episode_idx": 19, + "seed": 61, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 20, + "task_name": "CheesyBread", + "episode_idx": 20, + "seed": 62, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 21, + "task_name": "CheesyBread", + "episode_idx": 21, + "seed": 63, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 22, + "task_name": "CheesyBread", + "episode_idx": 22, + "seed": 64, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 27.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 23, + "task_name": "CheesyBread", + "episode_idx": 23, + "seed": 65, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 24, + "task_name": "CheesyBread", + "episode_idx": 24, + "seed": 66, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 28.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 25, + "task_name": "CheesyBread", + "episode_idx": 25, + "seed": 67, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 27.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 26, + "task_name": "CheesyBread", + "episode_idx": 26, + "seed": 68, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 27, + "task_name": "CheesyBread", + "episode_idx": 27, + "seed": 69, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 28, + "task_name": "CheesyBread", + "episode_idx": 28, + "seed": 70, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 29, + "task_name": "CheesyBread", + "episode_idx": 29, + "seed": 71, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 27.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 30, + "task_name": "CheesyBread", + "episode_idx": 30, + "seed": 72, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 31, + "task_name": "CheesyBread", + "episode_idx": 31, + "seed": 73, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 32, + "task_name": "CheesyBread", + "episode_idx": 32, + "seed": 74, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 33, + "task_name": "CheesyBread", + "episode_idx": 33, + "seed": 75, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 28.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 34, + "task_name": "CheesyBread", + "episode_idx": 34, + "seed": 76, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 27.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 35, + "task_name": "CheesyBread", + "episode_idx": 35, + "seed": 77, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 36, + "task_name": "CheesyBread", + "episode_idx": 36, + "seed": 78, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 28.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 37, + "task_name": "CheesyBread", + "episode_idx": 37, + "seed": 79, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 27.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 38, + "task_name": "CheesyBread", + "episode_idx": 38, + "seed": 80, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 39, + "task_name": "CheesyBread", + "episode_idx": 39, + "seed": 81, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 21.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 40, + "task_name": "CheesyBread", + "episode_idx": 40, + "seed": 82, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 41, + "task_name": "CheesyBread", + "episode_idx": 41, + "seed": 83, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 42, + "task_name": "CheesyBread", + "episode_idx": 42, + "seed": 84, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 43, + "task_name": "CheesyBread", + "episode_idx": 43, + "seed": 85, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 44, + "task_name": "CheesyBread", + "episode_idx": 44, + "seed": 86, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 45, + "task_name": "CheesyBread", + "episode_idx": 45, + "seed": 87, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 46, + "task_name": "CheesyBread", + "episode_idx": 46, + "seed": 88, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 26.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 47, + "task_name": "CheesyBread", + "episode_idx": 47, + "seed": 89, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 48, + "task_name": "CheesyBread", + "episode_idx": 48, + "seed": 90, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 49, + "task_name": "CheesyBread", + "episode_idx": 49, + "seed": 91, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 31.0, + "env_idx": 9, + "video_path": null + } +] \ No newline at end of file diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/CheesyBread/stats.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/CheesyBread/stats.json new file mode 100644 index 0000000000000000000000000000000000000000..76b286667268fc20ff77866b0a012c0427f017c5 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/CheesyBread/stats.json @@ -0,0 +1,606 @@ +{ + "num_episodes": 50, + "success_rate": 0.36, + "episodes": [ + { + "completion_order": 0, + "task_name": "CheesyBread", + "episode_idx": 0, + "seed": 42, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 1, + "task_name": "CheesyBread", + "episode_idx": 1, + "seed": 43, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 16.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 2, + "task_name": "CheesyBread", + "episode_idx": 2, + "seed": 44, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 25.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 3, + "task_name": "CheesyBread", + "episode_idx": 3, + "seed": 45, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 4, + "task_name": "CheesyBread", + "episode_idx": 4, + "seed": 46, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 5, + "task_name": "CheesyBread", + "episode_idx": 5, + "seed": 47, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 6, + "task_name": "CheesyBread", + "episode_idx": 6, + "seed": 48, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 7, + "task_name": "CheesyBread", + "episode_idx": 7, + "seed": 49, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 8, + "task_name": "CheesyBread", + "episode_idx": 8, + "seed": 50, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 27.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 9, + "task_name": "CheesyBread", + "episode_idx": 9, + "seed": 51, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 47, + "reward": 33.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 10, + "task_name": "CheesyBread", + "episode_idx": 10, + "seed": 52, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 11, + "task_name": "CheesyBread", + "episode_idx": 11, + "seed": 53, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 12, + "task_name": "CheesyBread", + "episode_idx": 12, + "seed": 54, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 29.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 13, + "task_name": "CheesyBread", + "episode_idx": 13, + "seed": 55, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 14, + "task_name": "CheesyBread", + "episode_idx": 14, + "seed": 56, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 15, + "task_name": "CheesyBread", + "episode_idx": 15, + "seed": 57, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 25.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 16, + "task_name": "CheesyBread", + "episode_idx": 16, + "seed": 58, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 17, + "task_name": "CheesyBread", + "episode_idx": 17, + "seed": 59, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 18, + "task_name": "CheesyBread", + "episode_idx": 18, + "seed": 60, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 33.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 19, + "task_name": "CheesyBread", + "episode_idx": 19, + "seed": 61, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 20, + "task_name": "CheesyBread", + "episode_idx": 20, + "seed": 62, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 21, + "task_name": "CheesyBread", + "episode_idx": 21, + "seed": 63, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 22, + "task_name": "CheesyBread", + "episode_idx": 22, + "seed": 64, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 27.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 23, + "task_name": "CheesyBread", + "episode_idx": 23, + "seed": 65, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 24, + "task_name": "CheesyBread", + "episode_idx": 24, + "seed": 66, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 28.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 25, + "task_name": "CheesyBread", + "episode_idx": 25, + "seed": 67, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 27.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 26, + "task_name": "CheesyBread", + "episode_idx": 26, + "seed": 68, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 27, + "task_name": "CheesyBread", + "episode_idx": 27, + "seed": 69, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 28, + "task_name": "CheesyBread", + "episode_idx": 28, + "seed": 70, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 29, + "task_name": "CheesyBread", + "episode_idx": 29, + "seed": 71, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 27.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 30, + "task_name": "CheesyBread", + "episode_idx": 30, + "seed": 72, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 31, + "task_name": "CheesyBread", + "episode_idx": 31, + "seed": 73, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 32, + "task_name": "CheesyBread", + "episode_idx": 32, + "seed": 74, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 33, + "task_name": "CheesyBread", + "episode_idx": 33, + "seed": 75, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 28.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 34, + "task_name": "CheesyBread", + "episode_idx": 34, + "seed": 76, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 27.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 35, + "task_name": "CheesyBread", + "episode_idx": 35, + "seed": 77, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 36, + "task_name": "CheesyBread", + "episode_idx": 36, + "seed": 78, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 28.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 37, + "task_name": "CheesyBread", + "episode_idx": 37, + "seed": 79, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 27.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 38, + "task_name": "CheesyBread", + "episode_idx": 38, + "seed": 80, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 39, + "task_name": "CheesyBread", + "episode_idx": 39, + "seed": 81, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 21.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 40, + "task_name": "CheesyBread", + "episode_idx": 40, + "seed": 82, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 41, + "task_name": "CheesyBread", + "episode_idx": 41, + "seed": 83, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 42, + "task_name": "CheesyBread", + "episode_idx": 42, + "seed": 84, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 43, + "task_name": "CheesyBread", + "episode_idx": 43, + "seed": 85, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 44, + "task_name": "CheesyBread", + "episode_idx": 44, + "seed": 86, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 45, + "task_name": "CheesyBread", + "episode_idx": 45, + "seed": 87, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 46, + "task_name": "CheesyBread", + "episode_idx": 46, + "seed": 88, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 26.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 47, + "task_name": "CheesyBread", + "episode_idx": 47, + "seed": 89, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 48, + "task_name": "CheesyBread", + "episode_idx": 48, + "seed": 90, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 49, + "task_name": "CheesyBread", + "episode_idx": 49, + "seed": 91, + "description": "Pick up the wedge of cheese and place it on the slice of bread to prepare a simple cheese on bread dish.", + "success": true, + "length": 48, + "reward": 31.0, + "env_idx": 9, + "video_path": null + } + ] +} \ No newline at end of file diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/CloseCabinet/episode_results.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/CloseCabinet/episode_results.json new file mode 100644 index 0000000000000000000000000000000000000000..d6e185d30aa42963768d1a7dd729a9f422d02217 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/eval_seed42_atomic39_nenv10_ep50_video/checkpoint-5000/evals/pretrain/CloseCabinet/episode_results.json @@ -0,0 +1,602 @@ +[ + { + "completion_order": 0, + "task_name": "CloseCabinet", + "episode_idx": 0, + "seed": 42, + "description": "Close the cabinet doors.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 1, + "task_name": "CloseCabinet", + "episode_idx": 1, + "seed": 43, + "description": "Close the cabinet doors.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 2, + "task_name": "CloseCabinet", + "episode_idx": 2, + "seed": 44, + "description": "Close the cabinet door.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 3, + "task_name": "CloseCabinet", + "episode_idx": 3, + "seed": 45, + "description": "Close the cabinet door.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 4, + "task_name": "CloseCabinet", + "episode_idx": 4, + "seed": 46, + "description": "Close the cabinet doors.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 5, + "task_name": "CloseCabinet", + "episode_idx": 5, + "seed": 47, + "description": "Close the cabinet doors.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 6, + "task_name": "CloseCabinet", + "episode_idx": 6, + "seed": 48, + "description": "Close the cabinet doors.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 7, + "task_name": "CloseCabinet", + "episode_idx": 7, + "seed": 49, + "description": "Close the cabinet door.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 8, + "task_name": "CloseCabinet", + "episode_idx": 8, + "seed": 50, + "description": "Close the cabinet doors.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 9, + "task_name": "CloseCabinet", + "episode_idx": 9, + "seed": 51, + "description": "Close the cabinet doors.", + "success": false, + "length": 47, + "reward": 0.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 10, + "task_name": "CloseCabinet", + "episode_idx": 10, + "seed": 52, + "description": "Close the cabinet door.", + "success": true, + "length": 48, + "reward": 37.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 11, + "task_name": "CloseCabinet", + "episode_idx": 11, + "seed": 53, + "description": "Close the cabinet door.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 12, + "task_name": "CloseCabinet", + "episode_idx": 12, + "seed": 54, + "description": "Close the cabinet doors", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 13, + "task_name": "CloseCabinet", + "episode_idx": 13, + "seed": 55, + "description": "Close the cabinet door.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 14, + "task_name": "CloseCabinet", + "episode_idx": 14, + "seed": 56, + "description": "Close the cabinet doors", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 15, + "task_name": "CloseCabinet", + "episode_idx": 15, + "seed": 57, + "description": "Close the cabinet doors", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 16, + "task_name": "CloseCabinet", + "episode_idx": 16, + "seed": 58, + "description": "Close the cabinet doors", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 17, + "task_name": "CloseCabinet", + "episode_idx": 17, + "seed": 59, + "description": "Close the cabinet doors", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 18, + "task_name": "CloseCabinet", + "episode_idx": 18, + "seed": 60, + "description": "Close the cabinet doors", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 19, + "task_name": "CloseCabinet", + "episode_idx": 19, + "seed": 61, + "description": "Close the cabinet door.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 20, + "task_name": "CloseCabinet", + "episode_idx": 20, + "seed": 62, + "description": "Close the cabinet doors.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 21, + "task_name": "CloseCabinet", + "episode_idx": 21, + "seed": 63, + "description": "Close the cabinet doors.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 22, + "task_name": "CloseCabinet", + "episode_idx": 22, + "seed": 64, + "description": "Close the cabinet doors.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 23, + "task_name": "CloseCabinet", + "episode_idx": 23, + "seed": 65, + "description": "Close the cabinet doors.", + "success": true, + "length": 48, + "reward": 17.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 24, + "task_name": "CloseCabinet", + "episode_idx": 24, + "seed": 66, + "description": "Close the cabinet doors.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 25, + "task_name": "CloseCabinet", + "episode_idx": 25, + "seed": 67, + "description": "Close the cabinet doors.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 26, + "task_name": "CloseCabinet", + "episode_idx": 26, + "seed": 68, + "description": "Close the cabinet doors.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 27, + "task_name": "CloseCabinet", + "episode_idx": 27, + "seed": 69, + "description": "Close the cabinet doors.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 28, + "task_name": "CloseCabinet", + "episode_idx": 28, + "seed": 70, + "description": "Close the cabinet door.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 29, + "task_name": "CloseCabinet", + "episode_idx": 29, + "seed": 71, + "description": "Close the cabinet door.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 30, + "task_name": "CloseCabinet", + "episode_idx": 30, + "seed": 72, + "description": "Close the cabinet doors.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 31, + "task_name": "CloseCabinet", + "episode_idx": 31, + "seed": 73, + "description": "Close the cabinet doors.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 32, + "task_name": "CloseCabinet", + "episode_idx": 32, + "seed": 74, + "description": "Close the cabinet doors.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 33, + "task_name": "CloseCabinet", + "episode_idx": 33, + "seed": 75, + "description": "Close the cabinet door.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 34, + "task_name": "CloseCabinet", + "episode_idx": 34, + "seed": 76, + "description": "Close the cabinet door.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 35, + "task_name": "CloseCabinet", + "episode_idx": 35, + "seed": 77, + "description": "Close the cabinet doors.", + "success": true, + "length": 48, + "reward": 10.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 36, + "task_name": "CloseCabinet", + "episode_idx": 36, + "seed": 78, + "description": "Close the cabinet doors.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 37, + "task_name": "CloseCabinet", + "episode_idx": 37, + "seed": 79, + "description": "Close the cabinet door.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 38, + "task_name": "CloseCabinet", + "episode_idx": 38, + "seed": 80, + "description": "Close the cabinet door.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 39, + "task_name": "CloseCabinet", + "episode_idx": 39, + "seed": 81, + "description": "Close the cabinet door.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 9, + "video_path": null + }, + { + "completion_order": 40, + "task_name": "CloseCabinet", + "episode_idx": 40, + "seed": 82, + "description": "Close the cabinet door.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 0, + "video_path": null + }, + { + "completion_order": 41, + "task_name": "CloseCabinet", + "episode_idx": 41, + "seed": 83, + "description": "Close the cabinet doors", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 1, + "video_path": null + }, + { + "completion_order": 42, + "task_name": "CloseCabinet", + "episode_idx": 42, + "seed": 84, + "description": "Close the cabinet door.", + "success": true, + "length": 48, + "reward": 38.0, + "env_idx": 2, + "video_path": null + }, + { + "completion_order": 43, + "task_name": "CloseCabinet", + "episode_idx": 43, + "seed": 85, + "description": "Close the cabinet doors", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 3, + "video_path": null + }, + { + "completion_order": 44, + "task_name": "CloseCabinet", + "episode_idx": 44, + "seed": 86, + "description": "Close the cabinet door.", + "success": true, + "length": 48, + "reward": 26.0, + "env_idx": 4, + "video_path": null + }, + { + "completion_order": 45, + "task_name": "CloseCabinet", + "episode_idx": 45, + "seed": 87, + "description": "Close the cabinet door.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 5, + "video_path": null + }, + { + "completion_order": 46, + "task_name": "CloseCabinet", + "episode_idx": 46, + "seed": 88, + "description": "Close the cabinet doors", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 6, + "video_path": null + }, + { + "completion_order": 47, + "task_name": "CloseCabinet", + "episode_idx": 47, + "seed": 89, + "description": "Close the cabinet door.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 7, + "video_path": null + }, + { + "completion_order": 48, + "task_name": "CloseCabinet", + "episode_idx": 48, + "seed": 90, + "description": "Close the cabinet door.", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 8, + "video_path": null + }, + { + "completion_order": 49, + "task_name": "CloseCabinet", + "episode_idx": 49, + "seed": 91, + "description": "Close the cabinet doors", + "success": false, + "length": 48, + "reward": 0.0, + "env_idx": 9, + "video_path": null + } +] \ No newline at end of file diff --git a/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/experiment_cfg/metadata.json b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/experiment_cfg/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..23dcfa871a88e56280f945357aef0296bd071711 --- /dev/null +++ b/FewShot/39tasks/5shot/FromBaseline26_Processingline/seed2/default/train_5shot_seed2_processingline_12k/experiment_cfg/metadata.json @@ -0,0 +1,431 @@ +{ + "new_embodiment": { + "statistics": { + "state": { + "base_position": { + "max": [ + 7.778976917266846, + 0.33403685688972473, + 0.7223060727119446 + ], + "min": [ + -4.821934223175049, + -6.785110950469971, + 0.6994727253913879 + ], + "mean": [ + 2.577927505176127, + -2.1525618086425604, + 0.7003541659101092 + ], + "std": [ + 1.6720365948940972, + 1.3768414767482617, + 0.0009102921174316669 + ], + "q01": [ + -0.7605057497562686, + -5.518671947805291, + 0.6999764703640262 + ], + "q99": [ + 6.038879433964306, + -0.5168299981147954, + 0.7029182461269848 + ] + }, + "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, + 0.24782371371751144, + 0.5971093380149053 + ], + "std": [ + 0.0, + 0.0, + 0.6642559499983499, + 0.3751891785184018 + ], + "q01": [ + 0.0, + 0.0, + -1.0, + 4.194193576416142e-06 + ], + "q99": [ + 0.0, + 0.0, + 1.0, + 1.0 + ] + }, + "end_effector_position_relative": { + "max": [ + 0.8848381638526917, + 0.8569459915161133, + 1.053884744644165 + ], + "min": [ + -0.5828549861907959, + -0.8486236333847046, + -0.3446122705936432 + ], + "mean": [ + 0.25006971579930243, + -0.005716176032514759, + 0.4426027721228362 + ], + "std": [ + 0.15838384751435522, + 0.25141504139800513, + 0.20646816234620438 + ], + "q01": [ + -0.017003721077380903, + -0.47738320165276543, + 0.1917563409753314 + ], + "q99": [ + 0.5401029199137674, + 0.4706646609268599, + 0.6965586537529516 + ] + }, + "end_effector_rotation_relative": { + "max": [ + 0.9999949932098389, + 0.9997354745864868, + 0.9900747537612915, + 0.9735639095306396 + ], + "min": [ + -0.9999914169311523, + -0.9999884366989136, + -0.9975369572639465, + 6.754255821306288e-08 + ], + "mean": [ + -0.22531628862537245, + 0.045669797538445966, + -0.06098055586474368, + 0.12292005493282424 + ], + "std": [ + 0.8506341437448487, + 0.2880763358322842, + 0.30816573781739026, + 0.16366165154486362 + ], + "q01": [ + -0.9957510593732032, + -0.5584741016891311, + -0.5371647381159943, + 0.0013784815838340492 + ], + "q99": [ + 0.9899375209218226, + 0.6327878884078038, + 0.5071539024531994, + 0.43555975763914345 + ] + }, + "gripper_qpos": { + "max": [ + 0.05289580300450325, + 0.00937709677964449 + ], + "min": [ + -0.011669117026031017, + -0.06307756155729294 + ], + "mean": [ + 0.03196506153371015, + -0.03229791582989581 + ], + "std": [ + 0.01171640503534049, + 0.01140016755425199 + ], + "q01": [ + 0.006290779996034005, + -0.04059717954551848 + ], + "q99": [ + 0.04038165838477291, + -0.006363076424932449 + ] + } + }, + "action": { + "base_motion": { + "max": [ + 1.0, + 1.0, + 1.0, + 0.0 + ], + "min": [ + -1.0, + -1.0, + -1.0, + 0.0 + ], + "mean": [ + 0.006112990930475949, + 0.002504955933349057, + -0.001639409958610569, + 0.0 + ], + "std": [ + 0.11478077265783658, + 0.1002662368589144, + 0.0807258815139021, + 0.0 + ], + "q01": [ + -0.16969970921613756, + -0.19919537649977875, + -0.15012139866371593, + 0.0 + ], + "q99": [ + 0.2466625453958827, + 0.25040842772679267, + 0.12777701094549537, + 0.0 + ] + }, + "control_mode": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.9253067479599785 + ], + "std": [ + 0.3792397014797737 + ], + "q01": [ + -1.0 + ], + "q99": [ + -0.1974093068624371 + ] + }, + "end_effector_position": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + -0.006754222051360652, + 0.0027854853184013867, + -0.08087572130330477 + ], + "std": [ + 0.45779905420789546, + 0.40578975219522756, + 0.4062694098132682 + ], + "q01": [ + -0.968295960290036, + -0.9240402337368848, + -0.8249359802202183 + ], + "q99": [ + 0.8046634915851651, + 0.9212243025654695, + 0.838433711437593 + ] + }, + "end_effector_rotation": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.008633749251857347, + -0.020406457524911718, + -0.0027639446362883006 + ], + "std": [ + 0.11393972872407104, + 0.13300186268009506, + 0.12192288302920105 + ], + "q01": [ + -0.2668579353776187, + -0.40299251206771697, + -0.34712957892172586 + ], + "q99": [ + 0.32748281264383927, + 0.3179028969292779, + 0.33459405510046736 + ] + }, + "gripper_close": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.3091635769453286 + ], + "std": [ + 0.9510122432876397 + ], + "q01": [ + -1.0 + ], + "q99": [ + 0.7643545615216344 + ] + } + } + }, + "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/FewShot/39tasks/5shot/launch_eval_39tasks_5shot_atomic39.concurrent.nohup.pid b/FewShot/39tasks/5shot/launch_eval_39tasks_5shot_atomic39.concurrent.nohup.pid new file mode 100644 index 0000000000000000000000000000000000000000..0a54028ac6800897e8212651b900a5218b09c6cf --- /dev/null +++ b/FewShot/39tasks/5shot/launch_eval_39tasks_5shot_atomic39.concurrent.nohup.pid @@ -0,0 +1 @@ +2594043